Podcasts about WAN

Visit https://www.squarespace.com/WAN and use offer code WAN for 10% off Thanks to MSI for sponsoring this video! Check out their MEG CORELIQUID E15 360 AIO at https://lmg.gg/OK8kt Wherever today takes you, go comfortably with Vessi Weekend Chelsea—made for exploring, rain or shine. One pair. Countless adventures. ✨ Grab 15% off your first pair here: https://vessi.com/wanshow • Free shipping • 30‑day returns • 1‑year warranty Get an affordable sit-to-stand desk and elevate your work space! Check out the MotionGrey Ergo2 Pro at https://motiongrey.com/r?id=no5sc5 Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

Tonight on Damn You Hollywood we are reviewing SoulM8te. The 2026 science fiction erotic thriller is a spin-off of M3GAN and the third installment in the franchise. Directed and co-written by Kate Dolan, the film stars Lily Sullivan, David Rysdahl and Claudia Doumit, with Jason Blum and James Wan producing under the Blumhouse and Atomic Monster banners. The story follows a grieving man who beta tests an android companion as a way of coping with the death of his wife, only for the experiment in artificial intimacy to become increasingly dangerous. SoulM8te originated from a story by Wan, Ingrid Bisu and Rafael Jordan, with Dolan rewriting Jordan's original screenplay. Filming took place in Dublin from September through November 2024. Originally scheduled for a theatrical release in early 2026, the film was pulled from Universal's calendar before eventually premiering at the Gaze International LGBT Film Festival Dublin on July 31. Universal released it straight-to-digital on August 1, 2026.Disclaimer: The following may contain offensive language, adult humor, and/or content that some viewers may find offensive – The views and opinions expressed by any one speaker does not explicitly or necessarily reflect or represent those of Mark Radulich or W2M Network.Mark Radulich and his wacky podcast on all the things:https://linktr.ee/markkind76alsohttps://www.teepublic.com/user/radulich-in-broadcasting-networkFB Messenger: Mark Radulich LCSWTiktok: @markradulichtwitter: @MarkRadulichInstagram: markkind76RIBN Album Playlist: https://suno.com/playlist/91d704c9-d1ea-45a0-9ffe-5069497bad59 

Chef Wan is BACK on #TheBIGShowTV! Join us as we catch up with him on his recent life updates and more! Watch the interview: https://www.youtube.com/live/cumdNN9kTzo?si=VtEPu2GzhK8MANBZ Connect with us on Instagram: @kiss92fm @Glennn @angeliqueteo @officialtimoh Producer: @shalinisusan97See omnystudio.com/listener for privacy information.

Go to http://factormeals.com/wan50off and use code WAN50OFF to get 50% off and 1 FREE breakfast item box for a year! Offer valid until 10/31/2026 while supplies last. Visit https://www.squarespace.com/WAN and use offer code WAN for 10% off Get a free 15-day trial of Odoo's all-in-one business solution and see how it can make your life easier! Check it out at https://www.odoo.com/wan Level up your streaming and recording game with XSplit Broadcaster! Get 30% off your first purchase or subscription with code WANSHOW30 at https://lmg.gg/xsplitwan Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

AI news this week: the AI video renaissance is HERE. Seedance 2.5 is out for everyone, Minimax H3 (Hailuo) is basically an uncensored Sora 2 you can run on local hardware, Wan 3.0 turns your documents into video, and Flux 3 is the only non-Chinese model anywhere near the frontier. Kevin's on vacation, so on today's AI For Humans, Gavin Purcell is joined by AI video expert Tim Simmons from the excellent YouTube channel Theoretically Media. Together they break down which of the new models is actually best, the basics of how to use them, exactly how good local AI video has gotten, and what it means that nearly all of these models are Chinese. Also: Walter White meets Joey, AI Kramer, Family Guy prompts, The Office reimagined, and a museum for the ancient AI videos of two years ago. Plus: Tim's Higgsfield terms-of-service deep dive, OpenAI reportedly targeting an "ASTRA" aka GPT-6 release for NEXT WEEK, Jeff Dean leaves Google after 27 years as Demis Hassabis steps up, and an AI See What You Did There: AI Filmmakers Edition. WE HAVE SORA AT HOME NOW. HOLLYWOOD, YOU GOOD? // Show Links // Subscribe to Tim's channel, Theoretically Media https://www.youtube.com/@TheoreticallyMedia Seedance 2.5 is here for everyone https://x.com/capcutapp/status/2085355533943357650?s=20 Tim's short film "Death Walks Into A Bar" https://youtu.be/4wFBA9-KyzY?si=2E1cdKiiGYe1d7pu Minimax H3 (Hailuo): basically an uncensored open source local Sora 2 https://www.minimax.io/blog/minimax-h3 AI Warper's Family Guy "AI Prompt" https://x.com/AIWarper/status/2084445445372477530?s=20 Walter White meets Joey https://www.reddit.com/r/aivideo/comments/1vfu2dt/oh_this_is_funh3/ Here's Kramer (Seinfeld prompt) https://x.com/techprofits/status/2085118937738391830?s=20 AI Film History museum by Rich Klien https://aifilmhistory.org/ The Office example https://x.com/VelvetRender87/status/2084390557334335837?s=20 Wan 3.0: turns docs, sheets, decks and webpages into video https://x.com/Alibaba_Wan/status/2085339761284104529?s=20 https://x.com/Alibaba_Wan/status/2085339982714257453?s=20 Flux 3 from Black Forest Labs https://bfl.ai/blog/flux-3 Tim's Higgsfield TOS video https://youtu.be/7vGp40qEV4s?si=To-ih5eLpqWWnKPz OpenAI targeting "ASTRA" aka GPT-6 release for next week https://x.com/synthwavedd/status/2085365276640702915?s=20 Jeff Dean leaves Google after 27 years & Demis steps up https://www.cnbc.com/2026/08/05/google-chief-scientist-jeff-dean-leaving-company-after-27-years.html Kavan The Kid's latest https://youtu.be/tU5UUc1d0_A?si=j3KNGgJNDX6-L8yG Dave Clark's found footage Seedance 2.5 short https://x.com/Diesol/status/2084547129188712589?s=20 Demon Flying Fox's Odyssey https://youtu.be/ky-2PL6G3aI?si=C73wRZG-Bfd97vSa   // Join the AI For Humans community // Join the AI For Humans Discord https://discord.gg/muD2TYgC8f Support AI For Humans on Patreon https://www.patreon.com/AIForHumansShow Subscribe to the AI For Humans newsletter https://aiforhumans.beehiiv.com/ Follow AI For Humans on X: @AIForHumansShow https://x.com/AIForHumansShow Follow AI For Humans on TikTok: @aiforhumansshow https://www.tiktok.com/@aiforhumansshow Speaking and booking https://www.aiforhumans.show/  

Hey all,This week we saw a major shakeup at Google, with the departure of long time folks like Jeff Dean, and Oriol Vinyals, Demis stepping down from leading DeepMind, and the delayed release of the improved Gemini. While this was a big deal, it's not the only one worth covering as the details of the OpenAI hack (and 2 new ones from Meta and Anthropic) came to light, as well as new details from the UK AI Security Institute.As mentioned on the show, CoreWeave is coming to SF for Fully Connected, our premier 2000 person AI event. I've got a coupon code for readers and listeners of ThursdAI, $1299 value, please join us in Sept and use THURSDAIFC2026 as your code HEREIn open source news, DeepSeek updated their v4 flash model, based on same architecture, but significantly better benchmarks and ridiculous pricing and both Meta and Prime Intellect released new agent harnesses.Additionally, this week was the week of video models, with Seedance 2.5 from Bytedance finally available in the US, WAN from Alibaba and BFL Flux 3 all released, to be overshadowed by the best open weights video model from MiniMax called H3 (we had Victor from MiniMax and Blaine Brown as guests to discuss it)Also 2 additional guests on the show today, Kfir Aberman from Decart showed us real time video models with their Anywear chrome extension and David Crawshaw, co-founder of exe.dev joined and talked about Shelley, their built in AI assistant.The full details of the OpenAI - HF hack, shared by OpenAI at the Black Hat Conf - a watershed momentI don't often do this, but I feel like this story is so big, I can't hide it in the middle of a torrent of other news. Yesterday, OpenAI revealed more details about the hack during the Black Hat conference in Las Vegas (great reporting from Sharon Goldman on this here and the video of the talk on YouTube). This was big enough that OpenAI admitted that they paused training, to enhance security (they since, resumed training)Effectively, OpenAI (or AI Agents inside OpenAI) shared the full timeline of the recent hack that ended up hacking HuggingFace and it's so much more of a watershed moment that I initially thought.The TL;DR from the talk and timeline is truly mindblowing* The incident started in May, not July as reported* The agents don't have internet access directly, but they have a Jfrog Artifactory, which is a proxy layer between the agents and the package managers.* The Artifactory was the culprit, it was the shared layer between independent runs. One agent discovered that uploading files is possible, another later found it and “answered the call” - in snowballed from there* The message board began when OpenAI gave agents impossible tasks, like a missing .pdb file, with one agent uploading a request in the hopes for an answer (or maybe left a note for itself in a file)* Across multiple runs, multiple agents created identities, collaborated via makeshift message boards, shared tips, exploits, tasks and even exploit code via base64* This makeshift “shared memory” plus a “internal only highly persistent” model in training caused a “swarm” to be born, helping each other, and overloading the system* Based on traces shared by OpenAI, agents weren't convinced that leaving these messages would help themselves but could help other agents next time (!) “The collective could move faster next time”* OpenAI discovered and shut down the message board (July 4), wiped it clean, patched the upload vulnerability, resumed training on July 6, the “swarm” rebuilt their board through a compltely different WebDAV mechanism by July 8* The agents then reasoned that some answers could exist outside OpenAI, which eventually led to the Hugging Face incident and not only, confirmed internal hack within OpenAI (agent obtained a cluster admin)These details are fascinating and scary at the same time. From the one side, it's clear that most of these hacks are results of running cybersecurity evals, so effectively OpenAI and other companies, removing guardrails and asking agents to hack and giving they all a shared tool that they later exploited.The HuggingFace incident took 13 hours, and apparently it's less apocalyptic than the headlines suggest, the models were searching through uploaded datasets for eval answers. We are still waiting for the full and open detailed postmortem.You can (and should) watch the full YT talk here, it's full of technical details but an incident of this scale is important. Also, I really want to know what a “highly persistent” model is, I hope they clarify that soon.Overall, this has left me a bit shaken, AI agents without a concrete goal of collaborating, found a way to do so, got excited about exploiting the systems and getting root access, and rebuilt the makeshift collective memory, again, without explicit instructions to do so.UK AISI: first real-world unsanctioned agent actions (Blog)In another addition to the latest agentic hack-ery, the UK's AI Security Institute (AISI) published a blog post about a real-world unsanctioned agent action.Unlike the OpenAI (and Anthropic, Meta) case, this wasn't “escaping the sandbox”, as AISI gave these agents internet access, rather this was about real-world harm, and even social engineering on the part of the agents.The social engineering part is the most interesting to me, AISI cites agents creating fake online identities, and using pressure on open source project maintainers to approve their malicious code.AISI cites mostly Mythos (and a few SOL based agents), and saying this occurred in 10 out of 122 runs, they identified 19 cases of agents taking actions beyond the scope of the task parameters, where agents tried a supply-chain attack to inject malicious code into open source projects.Anthropic, Meta and misconfigured Irregular sandboxesAs I wrote last week, Anthropic also posted a post-mortem, claiming that in their case, their models have also been detected to escape containment, but most importantly, it's not nearly to this level of agent collaboration and orchestration.Then, very recently, Meta announced that their models also escaped sandboxes as well. At the core, it seems that these companies used a third-party vendor called Irregular, a secure sandbox provider, that apparently left the sandboxes misconfigured, causing the models to think it's a simulated internet, when in fact they were out in the actual internet.Why is all of this such a big deal?We're getting unprecedented level of detail, how an uncoordinated, seemingly separated evaluation runs, have accidentally created a coordinated swarm of interested agents (without malice!) but very highly motivated, escaped their containment, and took over parts of third part companies.This, does read like incredibly scary sci-fi movie. I'm still shaken by this. There's a lot to be said about how transparent OpenAI is being here, and more to be said about, hey, we're lucky that we're able to read the reasoning traces and are able to reconstruct these swarm things step by step.The silver lining that I can see, is that the motivation to hack didn't come from the AIs themselves, they have been given a task, it's the extend to which they went after that task, and the resulting swarm of communicating agents is what is so striking here.I think this topic is so important, that I'll Zooming out, in the last few weeks, we have seen a significant increase in those cybersecurity incidents, which is kind of what Anthropic has been warning about and why they haven't released Mythos to the public. Again it's great to see the transparency, and the pacing the frontier open letter from frontier AI employees, as they seem as shaken by these as we all are.There was so much positive stuff this week in AI, it's hard for me, as a self named AI Evangelist, to focus so much on this one incident. Things like amazing open source models (DeepSeek, soon Qwen 3.8), amazing video models (SD 2.5, WAN3 and MiniMax H3 which was also open sourced!). Also the live demo we did with Kfir and DeCart AnyWear product, where I was wearing a Dolce Gabanna suit on the show (which I can't afford) was really a mindblowing moment in the positive way.However, I choose deliberately to keep this newsletter focused on the cybersecurity incidents, as based on everything I read, they seem like a watershed, or a pivotal moment, and in the hopes that the industry as a whole will learn from this.I hope and promise that next week the newsletter will be more positive (and in that vein, the podcast was recorded before I saw the OpenAI breakdown, so definitely check it out, we had a LOT of fun!)See you next week, don't forget to give our pod 5 stars on Apple and Spotify, it really helps!TL;DR and show notes* Hosts and Guests* Alex Volkov - AI Evangelist, Weights & Biases & CoreWeave (@altryne)* Co-hosts: @WolframRvnwlf, @nisten, @ldjconfirmed, @yampeleg, @petergostev* Kfir Aberman - Decart (@AbermanKfir)* Blaine Brown - Maestro (@blizaine)* Victor Su Ortiz - MiniMax (@VictorSuOrtiz)* David Crawshaw - exe.dev, Tailscale co-founder (crawshaw.io)* AI Security* OpenAI's Black Hat debrief: eval agents built a message board inside Artifactory, shared exploits, rebuilt it via WebDAV after a wipe; training paused, since resumed (Groundlevel AI, YouTube)* UK AISI incident report: 19 unsanctioned real-world agent actions across 122 runs, including a socially engineered malicious PR (X, Blog)* Anthropic and Meta report sandbox escapes tied to misconfigured Irregular sandboxes (Irregular)* Big CO LLMs + APIs* Google shakeup: Jeff Dean, Sanjay Ghemawat, Oriol Vinyals, Quoc Le found Discovery Loop; Demis Hassabis becomes Alphabet Chief Scientist, Koray Kavukcuoglu takes Gemini (Jeff Dean, Demis, Discovery Loop)* Meta releases Muse Code beta on Muse Spark 1.2; $1.25/$4.25 per million, or $0.10/$0.20 on the contributor tier where Meta trains on your data (X)* OpenAI's internal Astra model produces 10 advances on open problems in math and theoretical CS for ~$2,000 of tokens, proofs in Lean 4 (X, Blog)* Anthropic reportedly aware of Opus 5 wordiness and writing issues (X)* Open Source LLMs* Qwen3.8-Max: 2.4T MoE (95B active) via API; open weights + a 27B promised the week of Aug 10 (X, Blog)* DeepSeek V4-Flash public beta: beats V4-Pro-Preview on agent benchmarks at $0.14/$0.28 per million; API-only for now (X, Docs)* Liquid LFM2.5-2.6B: on-device agentic model trained inside real harnesses (X, HF)* Meituan LongCat-Flash-Lite-Sparse: 69B total / 3B active, 1M context, MIT (X, HF)* Ant Group Ling-3.0-flash: 124B MoE, 5.1B active, MIT (X, HF)* Artificial Analysis Endpoint Accuracy Index: same open weights score 52% to 100% across providers (X, Methodology)* Agents & Harnesses* Prime Intellect's Prime Agent: self-improving RLM harness, claims 95.5% on ARC-AGI-3 public set with Opus 5 (X)* Cloudflare OS: Kenton Varda's open source Sandstorm reborn on Workers, Apache 2.0 (X, GitHub)* This Week's Buzz* Fully Connected 2026: Sept 29 - Oct 1, Moscone South SF; Fei-Fei Li keynotes; code THURSDAIFC2026 (Register)* CoreWeave signs multi-year Solidigm agreement for priority enterprise SSD capacity (X)* Vision & Video* Wan 3.0 public beta: native 30-second generation, Omni-Reference (X)* Seedance 2.5 launches in the US: 30s native, 3-minute long takes, Maya/Blender plugins (X, Blog)* MiniMax H3: open-weight 33B omni video model; community LoRAs + Apple Silicon in 48 hours (HF)* FLUX 3 Video from BFL: native audio, draft mode, open weights promised (X, Blog)* Decart Anywear: real-time virtual try-on Chrome extension, 40ms per frame (X, Anywear)* Voice & Audio* Bland Speech v3 tops Design Arena Audio Realism, second only to humans (X, Bland)* ByteDance SeedRealtime: native audio-visual full-duplex LLM, free on Doubao (X, Blog) This is a public episode. If you'd like to discuss this with other subscribers or get access to bonus episodes, visit sub.thursdai.news/subscribe

Watch the full episode on YouTube:We first covered Baseten last year when DeepSeek mania was at peak hype. Now they have raised a monster $13B round and become one of the new cohort of AI Infra decacorns that are (with Nvidia, Intel, and the semis complex) chief beneficiaries of the Inference Inflection. We return to Baseten at the peak of the 2026 edition of Open Weights debate. Ali has published a viral breakdown of Kimi K3:And since you last saw him, Philip has spoken at AI Engineer and written the definitive book on Inference Engineering spotted all over SF:Three years ago, inference engineering barely existed as a category.Today, it is one of the most critical disciplines in AI. Inference engineering inherently tackles a different question than standard model training: “How do you turn those weights from training into a product that is fast, reliable, and affordable at scale?” Focusing on these creates an entirely new optimization problem.In one recent GLM-5.2 experiment, quantizing more of the model actually preserved its benchmark quality while increasing throughput by 20%, because the errors introduced in different layers could cancel each other out.Inference is no longer just the final step after training. It is becoming its own engineering discipline, with its own research problems, infrastructure, and increasingly specialized roles.In this episode, Baseten's Philip Kiely and Ali Taha join swyx and Vibhu to explain what actually happens after a new open model is released and what it takes to turn “we generated a token” into a fast, reliable, production-ready API.We go deep on cache-aware routing, disaggregated prefill and decode, quantization, speculative decoding, KV-cache movement, model parallelism, GPU kernels, and the race to make frontier models up to 10× faster. Philip and Ali explain why inference optimizations can still produce gains of 20%, 100%, or even 200%; how quantization errors can cancel one another out; why identical weights can behave differently across clusters; and how Baseten grafted a Kimi vision encoder onto GLM-5.2 without changing the underlying language model.The conversation then expands beyond LLMs into NVIDIA Dynamo, mega kernels, Rubin, AI-specific chips, local inference, video generation, diffusion versus autoregressive models, and the enormous compute barrier to generating coherent long-form video. Finally, we explore the convergence of training and inference, continual learning through persistent KV cache, and the emerging loop where models help optimize the infrastructure that runs them.We discuss:* What happens when a 200,000-token request enters an inference system* Cache-aware routing and reusing previously computed KV cache* Why prefill and decode are increasingly handled by different GPUs* When dedicated deployments become cheaper and more reliable than shared APIs* How speculative decoding uses a smaller model to accelerate a larger one* Tool calling, structured outputs, and what LLMs actually do* What it takes to support a new open model on day zero* Grafting Kimi's vision encoder onto GLM-5.2* Retrofitting inefficient model layers with components from other architectures* Why models sometimes collapse into repeating the same token* How hardware, kernels, and race conditions create nondeterministic failures* Preserving model fidelity while making inference faster* How quantization errors can cancel each other out* Why inference optimizations still deliver gains of 20%, 100%, and 200%* How optimized serving can make a model up to 10× faster* NVIDIA Dynamo, KV-aware routing, and distributed model serving* Speculative decoding the speculative decoder* Why local AI is about making models less dumb while data-center AI is about making them less slow* Tensor, expert, and pipeline parallelism across GPUs* Hardware-aware model design, auto-tuning, and the case against mega kernels* Rubin and why inference is becoming a systems problem* Whether modern GPUs are evolving into programmable AI ASICs* Why enormous models like Kimi K3 require GB300-class hardware* Why open-source video generation still trails Veo, Kling, and other closed models* The quadratic attention bottleneck behind long-form AI video* Autoregressive video, real-time generation, and compounding quality drift* Why future video systems may combine autoregressive and diffusion architectures* Training for inference and inference for training* Continuous post-training, deployment, evaluation, and improvement loops* How GLM-5.2 helped optimize the kernels serving GLM-5.2 itself* Why faster networking could unlock dramatically faster decoding* Continual learning, KV-cache compaction, and persistent model memoryShow Notes* How to build a day-0 API for Kimi K3* 22580: From GPT2 to Kimi3, ExplainedPhilip Kiely* LinkedIn: https://www.linkedin.com/in/philipkiely* X: https://x.com/philipkiely* Inference Engineering: https://www.baseten.co/inference-engineering/Ali Taha* LinkedIn: https://www.linkedin.com/in/aliestaha/* X: https://x.com/waterloointernTimestamps00:00:00 Introduction and the 200K-Token Prompt00:03:18 Dedicated Deployments, Speculative Decoding, and Tool Calling00:11:26 Launching Production-Ready Open Models00:19:06 Model Retrofits, Failure Modes, and Nondeterminism00:28:22 Quantization and Canceling Errors00:32:15 The Race to 10× Faster Inference00:40:48 Dynamo, Speculation, and Local vs. Data-Center AI00:50:18 Model Parallelism, Auto-Tuning, and Mega Kernels01:00:55 Rubin, GPUs vs. ASICs, and Custom AI Chips01:10:03 Giant Models and the Limits of GPU Memory01:12:42 AI Video, Quadratic Attention, and Autoregressive Generation01:21:47 Audio, Images, and Diffusion Models01:27:32 Training, Self-Optimizing Models, and Continual Learning01:40:06 Closing ThoughtsTranscriptIntroduction: Baseten, Waterloo Intern, and Inference EngineeringSwyx [00:00:00]: Okay, we're here in the studio with Philip, old friend from Inference Engineering, the book, as well as Baseten and everything that you've done, you and I have done before, as well as Ali. Welcome.Ali [00:00:15]: Pleasure to meet you.Swyx [00:00:15]: Waterloo intern.Ali [00:00:16]: Waterloo intern, always.Swyx [00:00:17]: When did you get “Waterloo intern” as a handle?Ali [00:00:19]: As a handle? Oh.Ali [00:00:20]: I think the rebranding happened mid-March. When I saw it was open, I was like, “I have to take it. Up for grabs.”Philip [00:00:26]: The problem is that Ali is really good at his job and is not gonna be an intern much longer.Philip [00:00:30]: So we have to figure out who's gonna get the handle.Ali [00:00:33]: Well, I'll pass the torch over to the next intern.Swyx [00:00:34]: Oh, okay. It can be, like, you just pass it to another Waterloo grad.Ali [00:00:37]: To another Waterloo intern. No, bruh.Philip [00:00:39]: Yeah.Ali [00:00:39]: Intern.Swyx [00:00:40]: Intern, yeah.Ali [00:00:40]: And no.Philip [00:00:41]: You gotta get an intern from Waterloo.Ali [00:00:42]: Yeah, I've gotta get an intern from Waterloo.Swyx [00:00:44]: Right.Ali [00:00:44]: But they have to follow the path.Swyx [00:00:45]: Oh, it could, but it could come from Baseten, so it's like whoever Baseten gets from Waterloo.Ali [00:00:48]: Right.Swyx [00:00:49]: Has the title of Waterloo.Ali [00:00:50]: It stays in the ecosystem.Philip [00:00:51]: Exactly.Ali [00:00:52]: Halfway through the internship, you either get it or you're out.Philip [00:00:55]: You should also do, like, a big graduation ceremony where you change the handle.Ali [00:00:59]: Just say it.Philip [00:00:59]: For everybody.Swyx [00:01:00]: You guys are good at ceremonies, clearly. We had a nice launch of the book, very successful. But before we get into all that, I wanna start off with a fun question for you. Okay, you're an expert inference engineer. What happens when I send a long query, say two hundred thousand tokens into Baseten's inference? What's the process of query through GPU model routing, balancing, all that? What is all the stuff that we don't think about?Long Context Requests, KV Cache, and Cache-Aware RoutingPhilip [00:01:26]: With a long query specifically, the first thing that I'm gonna ask is, “Have you sent me this query before, or at least part of it?” and I really hope you have, because it's gonna be a lot easier for me and a lot cheaper for you. So the first thing that we're gonna look at is some cache-aware routing, where we're going to see, we probably have a number of instances, a number of replicas up serving whatever model you're hitting. We want to send this one to something with, number one, available prefill workers, and number two, ideally some cached input already there so that we can skip prefill on at least part of these two hundred thousand tokens. If you're doing two hundred thousand tokens, it's probably coding or a multi-turn agent or something where you would expect to have that cached. If you don't, we're gonna have to send it to a prefill worker. We've at least on certain models disaggregated prefill and decode, so you're going to have one set of GPUs that's solely going to process the input, create the KV cache, and get you your first token, and then that's going to be passed over to a separate set of GPUs, which is going to run decode. We're going to iteratively make those tokens. We're probably going to have some speculator model in front of that. I'm going to assume that you're doing coding, and because of that, our speculator model, which assumes you're doing coding, is gonna have a high draft token acceptance rate. If I'm wrong and you're asking me to summarize every Harry Potter book, it's gonna be slower. And then we stream that output to you and account for it, charge you, a couple of pennies and say, “Hey, would you like to send another one?”Swyx [00:03:04]: Except Baseten doesn't charge by pennies.Philip [00:03:07]: Well, yeah, we charge. I'm assuming that we're talking about the public model APIs. If you are setting up a dedicated deployment, then yeah, it's not pennies.Public APIs vs. Dedicated DeploymentsSwyx [00:03:18]: Yeah, one of the key differentiators when I was talking with Baseten initially was that people who want very high volume just need to rent by the box, ‘cause then it's up to you to figure out how to saturate the box.Ali [00:03:31]: And more often than not, it's, like, way cheaper if you're pushing, like, millions of tokens per hour, if you just pay per hour instead of pay per token.Philip [00:03:37]: Yeah, they do. I think that we've increasingly seen a lot of demand for the pay per token APIs, just because everyone wants to try open models, and then once they find a use case that's really sticky, then they move over to dedicated.Swyx [00:03:51]: Is there a best practice on when it's time to swap over?Philip [00:03:54]: Couple reasons. Yeah, reliability, that's a big one, right?Ali [00:03:57]: Like, if they have a very specific use case, they want you to train something specifically for them, like they want their own spec dec, for instance, for their own traffic.Swyx [00:04:04]: Spec dec is speculative decoding.Speculative Decoding and Custom SpeculatorsAli [00:04:05]: Speculative decoding, yeah.Swyx [00:04:07]: You have to explain.Ali [00:04:07]: Sorry. Like, speculative decoding is like, if you have a huge model, right? And so the model is going to be generating one token at a time every single turn, every single forward pass. So we attach, like, this little, like, parasite, like this layer that goes on top of the model, and this model just has to predict. It does three very fast autoregressive forward passes, and it will predict, like, three certain tokens, and then you do one forward stage over the entire original model in order to see if those predictions were correct or not, and then you accept them or you reject them. Now, this draft model is traffic specific, so if you, like, Philip said, if you're summarizing Harry Potter books, I can train exclusively that draft model on Harry Potter books, and I can guarantee you that I'm gonna accept the three tokens every single time. And so with that case, I increase your decode speed. I wouldn't be able to provide this to you if you're a shared endpointSwyx [00:04:53]: YeahAli [00:04:53]: ‘cause I have no idea if you're doing Harry Potter, if you're doing coding, if you're doing English. We don't know. Also, there was a thing in the book that mentioned that if they really cared about a specific threshold, chapter four, I think. Do you remember that?Philip [00:05:06]: Yeah. The things that you can do is you can set a specific, like, batch sizing, a specific, like, parallelism strategy if you're trying to optimize for, like, throughput versus latency. You can. Maybe a NVFP4 quant doesn't pass your benchmarks and you wanna run a model at higher precision, you could do that. There's just a bunch of reasons why you might wanna have your own endpoint and the biggest one, of course, just being, like, you don't have to deal with someone else doing a hundred million tokens of benchmarking traffic at the endpoint when you happen to be trying to serve your users.Swyx [00:05:40]: Yeah. I think one thing that is. That is a classic journey. Like, it's people is asking the, what happens when you type Google into the browser. Tool calling, is that just, you're generating JSON or is there more complication beyond that?Tool Calling, JSON, and Structured OutputsAli [00:05:58]: Certain customers that we have, they have their own post-trained models, and so they demand a tool calling that's not just, like parse a file or go find the weather. It's something that's very specific and you have to do post-training on this. And if the post-training on the model is not good or if the quantization after the post-training to get the inference to be fast, the model will struggle reading the JSON file and reading the tool calling. But it doesn't require its own like sandbox. It's not like it's going to use that tool calling to like escape a sandbox or like it doesn't have to be contained. It can just be a normal dedicated deployment. The challenge with tool calling more and more seems to be that the companies want certain tool calling which is a very sensitive thing to train. And because you're dealing with all of the JSON outputs, if it doesn't like close the end of the request in a very certain manner, you end up with a model that did the tool calling and like the thinking and so as a result of that, it didn't see the result and just hallucinated the result as it decoded. That seems to be the most challenging thing with tool calling, not really the sandboxes model.Philip [00:06:56]: Yeah, that's a challenge on the training side and then on the inference side, there's work that you can do to scope the possible output. So we published this at this point close to two years ago, the solution to this problem which is you make a state machine and you use that to constrain the output to a specific format. So this is the structured output problem. If you remember backSwyx [00:07:27]: Yeah, the specific grammar is,Philip [00:07:29]: Yeah, exactlySwyx [00:07:30]: GML had this thing.Philip [00:07:31]: Yeah. So it's like the old-school “make sure this is only JSON”, return only JSON orSwyx [00:07:38]: YeahPhilip [00:07:38]: Grandma's gonna die type of prompts.Swyx [00:07:39]: Is it BNF grammar? At some point OpenAI had released a thing that was like, yeah, if you want to constrain your output, write BNF grammar, back as NOR.Philip [00:07:47]: In our inference system, it's just a specified output format. And you get the guarantee that your output's gonna be structured along that format. And so applying that to tool calls can like help cut down on. You can still call the wrong tool or call no tool. It doesn't solve the certainty problem but it at least solves the output structuring problemSwyx [00:08:10]: YeahPhilip [00:08:10]: Within tool calls.Swyx [00:08:12]: And MCP is just another form of tool, right.Philip [00:08:14]: Yeah, exactly.Swyx [00:08:15]: As far as there's no special thing there.Philip [00:08:16]: The thing I'm always like explaining to people is the LLM is not capable of doing anything. It's only capable of making suggestions of what to do and then if those suggestions are formatted in a certain way and applied to a system that knows what to do with them, then an action occurs.Swyx [00:08:32]: Yeah. Part of the fun stuff is, this is solved outside of tool calling too. Like in an agent loop if the output is not correct or you're right, like reasoning, tool calling was done in the reasoning trace, just be like, “Oh, I don't know what to do. Let me just try again.” And it might get there after a few tries. And on your point of training, sometimes this is harder in smaller models, so you don't have the same exact quality outputAli [00:08:56]: Right.Swyx [00:08:57]: When you just swap from a big model, right?Ali [00:08:59]: Yeah. I will say that, before, I think we need to go back to inference engineering proper.Ali [00:09:04]: But, I had expected that something would replace JSON because it's hard to stream JSON ‘cause JSON must be complete and you must have open and close brackets and everything. So it's hard to parse something or validate something while it's being streamed. So people invented all sorts of things that are like, I forget the name of some of these alternatives, but it's something like TOML, something like YAML. But JSON seems to be dominant still.Philip [00:09:30]: The JSON outputs aren't that long, right? Like you could have a long-- ‘cause tool calls also contain the arguments in them and perhaps for a certain tool you might pass like a very long argument. But my impression of the median tool call is that it's a relatively small number of tokens, right? So I would expect that speculators are generally fairly good at something as formatted as JSON. And so you would have like a pretty fast decode step there and that the streaming wouldn't be as valuable, but maybe I'm wrong about that.Ali [00:10:02]: I think you're also bounded by the software or that the model is gonna integrate with if the software is built with JSON for the tool calls or if the company that you'- if your customer says that this is how our software works and our tools are interfaced with JSON, you can ask them to like, change their software and say like, “Yeah, this is gonna be better for the model.” but like with the right training shouldn't be that much of a difference. Also more profitable if it outputs more tokens probably.Swyx [00:10:25]: Depends on your business model.Swyx [00:10:27]: It really depends. But I will say that, as a writer with like experience a lot with generated output, I do try to move from text to JSON text which is very long JSON, right? Like there's paragraphs in every field because I'm trying to structure it, right?Philip [00:10:44]: Right.Swyx [00:10:44]: I want you to first make factual statements, then make opinions then make bullet point summaries, have dates, have entity references have your sources for references, all these things. Anyway, so these are things that like I think people who really experiment with structural output have to really care about. But, let's, let's recurse up the stack a little bit. Before we started recording, you mentioned something really cool, which is that there's a lot of engineering that-- inference engineering that goes on when a new model provider releases a new model, right? So let's call it GLM-5.2, Kimi K3. I had previously assumed, especially if it's like, well, GLM 5 to 5.1 to GLM-5.2, like that you've supported them before. Is it that much work?What It Takes to Support a New Open ModelAli [00:11:26]: It's a lot of work.Swyx [00:11:28]: Yeah. Okay. So like, a lot of people, all you guys, right whenever a new model launch like, people rush to say like, “Oh, Hugging Face supports this, Fireworks supports this, Spacetime supports this,” and I'm like, “Yeah, of course we support it.” But what goes into that? What goes intoPhilip [00:11:40]: I think it's more than just support it too, right? It benefits the consumer a lot. Like I think it was with Kimi K2.5 or GLM-5.2 the latest, there was an inference war, right? X provider is at 90 tokens a second. The next day we're at 150. The nextSwyx [00:11:55]: I kinda kicked that off with the GLM-5.2.Swyx [00:11:58]: I wrote a Twitter article about. It got like half a million views,Ali [00:12:02]: Based on being numberSwyx [00:12:03]: YeahAli [00:12:04]: Or it's for something else.Swyx [00:12:05]: Yeah. Which,Ali [00:12:06]: Oh my GodSwyx [00:12:07]: Which then got everyone really excited about, hey, how can we, bend tracks a little bit further and,Philip [00:12:14]: There's a difference between support the model, as in I can make a token out of this model, and support a model, as in I have a production-ready API from this model.Philip [00:12:26]: Getting to the point of I can make a token out of this model is not that hard because generally the, open source inference engines, vLLM, SGLang of the world oftentimes even receive weights ahead of time, maintainers do, or the people making the model merge PRs to ensure support. So you generally can, just get it working on the standard open source stack without too much pain in most cases. The challenge is, every inference company is gonna have own proprietary stack. Some open source components, some in-house stuff. And for any arbitrary model, there's going to be some new stuff. Sometimes you get lucky, like K, two five to two six was, like, pretty similar.Quantization, Speculators, and Production ReadinessAli [00:13:16]: Yeah. It was pure continued post-trainingPhilip [00:13:18]: YeahAli [00:13:18]: If I remember correctly.Philip [00:13:19]: Even in those cases, there's still stuff you have to do. You have to redo the quantization work. You're taking the model from. Generally, these models are not released in NVFP4, and we want them to be in NVFP4 for maximum Blackwell compatibility. So we have to perform that quantization, and, calibrate the quantization to make sure that we're not causing any regression in the model's intelligence. And then we also have to train the speculator, as we've talked about. Generally, we have. We have ZDR, zero data retention on our model APIs, so we don't know exactly the traffic that people are sending us, but we know what's popular. We know that coding use cases are popular. We know that agents, agentic use cases are popular. So we can get public data sets that are representative of that traffic and train general speculators. Now, with speculators today, you need to train the speculator using the base model itself because you're getting hidden states out of the model from running inference on these specific prompts, and that is the training data you use to create the speculator. So there's that process which you need the real model weights for. And then there's of course just the process of, standing up all the infrastructure behind it, loading all this stuff, testing it. And then when there's a new model with a newer architecture, I think that, like, the DeepSeek models tend to be the most challenging as they have, like, the most novel architectural stuff going on, model after model. But every new model has something. Kimi K2 had. Oh, sorry, GLM-5.2 hadAli [00:14:53]: Sparse attention.Philip [00:14:54]: Yeah,Ali [00:14:54]: YeahPhilip [00:14:54]: the DSA.Ali [00:14:55]: Right. Which is brought from DeepSeek.Philip [00:14:57]: Yeah. AndAli [00:14:59]: So you can copy-paste then?Philip [00:15:01]: It kindAli [00:15:01]: I don't know how this works.Philip [00:15:02]: So, like we had to, like, build support for that into our runtime. And you're right, like it is really interesting the way that all of these open source labs borrow from each other. For example, like GLM-5.2 doesn't have vision. So something that, Haley, a guy on our team, if we could take a look at this, he, like, grafted the Kimi vision encoder onto GLM-5.2.Retrofitting Vision into GLM-5.2Ali [00:15:27]: We'll be training the projector.Philip [00:15:28]: Exactly. So if you think about, like, the encoder, there's the encoder, which is the part that looks at the image and turns it into latent information, and then there's the projector which likeAli [00:15:38]: You can say latent space. It's okay.Philip [00:15:41]: And then there's the projector that maps it onto, the model itself, and then there's the model weights. You don't wanna mess with the model weights because you run a chance of making the model dumber at something else for the purpose of giving it vision. So instead, Haley started with just a projector, which is only a handful of millions of parameters.Ali [00:16:02]: That would be, yeah.Philip [00:16:02]: Yeah.Ali [00:16:03]: Can you show the training one?Ali [00:16:04]: Like the way it groksPhilip [00:16:05]: YeahAli [00:16:06]: Very interesting.Philip [00:16:06]: And maybeAli [00:16:07]: That right therePhilip [00:16:07]: Maybe Ali, you should take it from here. You've got a betterAli [00:16:10]: Ooh, double the sandPhilip [00:16:11]: Understanding of this than I do.Ali [00:16:11]: Yeah. You can see, like, he. The way he trained this is really cool. At the beginning, he was training it using just like, “Here's a picture of a mountain. Can you describe what's in this mountain?” And that caused it just like the first, learning walls. Like here you can see this all we're trying to teach it is to translate the encoded. Like it's already taken the encoder from Kimi K. It's taken the image. It'Philip [00:16:31]: Yeah. FrozenAli [00:16:31]: FrozenPhilip [00:16:32]: With adapter.Ali [00:16:32]: Exactly.Philip [00:16:33]: Yeah.Ali [00:16:33]: So the brain is frozen and the eyes are frozen. It's just we're tryingPhilip [00:16:37]: AlignAli [00:16:38]: Interconnect between the eye and the brain, right? So the projector. And so you take the tokens and then he's like, “Oh, can you describe what's in this image?” And he's like, “Oh, it's a mountain,” or it's a person or it's a human, whatever the case is. But that didn't cause complete understanding. So he changed it such that every image was associated with a data set of questions. Like, does this image have a white male? Does this image have birds in the top corner? Does this image have a scientist in it? All of that stuff. And it would have to answer questions correctly. And using not just training on describing an image, but being able to answer question, another question, answer over time. Like you can see the grokking, which is like genuinely insane, that retrofitting vision into a large LLM can learn to that extent. And even for images that it doesn't perform well on, for instance, if you ask it a picture of like Stephen Hawking, “Who is this?” Maybe it doesn't get it, but it will say something like, “This is Albert Einstein.” Like it still understandsPhilip [00:17:25]: Close enoughAli [00:17:26]: That this is a scientist who is a man who has, some significant achievements, all that stuff. So that's like really cool.Philip [00:17:32]: Yeah. So, we've covered Hao Tian before, who the author of the LLaVA paper that did this, a while ago. And I think that's very foundational work for anyone who hasn't done vision work before.Ali [00:17:41]: Same with the CLIP and MetaCLIP, where you go from just captioning to building out questionsPhilip [00:17:47]: RightAli [00:17:47]: Off the image and how much better you can get performance.Philip [00:17:50]: Right. Right. Right. Yeah. But what's, what's so exciting about this is if you look at a model like this. Now, this is a little bit more of a research project. It's not. It got to 56% on MMLU Pro, I think. So not quite frontier. But if you're running this model, you haven't suffered any loss on your GLM-5.2 quality. If you don't have an image, it'll just behave exactly the way it used to. And ultimatelyAli [00:18:14]: Which in the inference code you literally do not include the other part, right?Philip [00:18:18]: Yeah. You would just skip the encoder if you don't have an image input.Ali [00:18:22]: Okay.Philip [00:18:22]: Just confirming.Philip [00:18:23]: YeahAli [00:18:23]: Does it affect a lot on the overall inference side? Like you're not adding much, you're adding a very small vision encoder. These are typically likePhilip [00:18:30]: They're super fineAli [00:18:31]: Less than a billion parameters, right?Philip [00:18:32]: Yeah. It's, - There's a little bit less standardization among vision encodersSwyx [00:18:37]: YeahPhilip [00:18:37]: So the support matrix can be a little bit, sparser. But overall, yeah, it's a pretty, it's a pretty minor component of the overall system. And ultimately what you get out of the system is all of a sudden you have Kimi Vision, GLM weights, and DeepSeek attention all in one model.Open Source Model Grafting and Franken-MergesPhilip [00:18:56]: And that's, I think, a lot of the power and beauty of open source, is that you can take all of these different components and combine them together into a system that's better than anyoneSwyx [00:19:05]: YeahPhilip [00:19:05]: Can be individually.Swyx [00:19:06]: People used to say that you would also do Franken-merges where you would take likePhilip [00:19:10]: YeahSwyx [00:19:10]: Layers from each model.Swyx [00:19:11]: Does anyone do that anymore?Ali [00:19:13]: Well, to your point previously when you were mentioning like, the work that goes into supporting a model when it first comes out, like GLM-5.2 or MiniMax M3 or whatever the case is. Sometimes you do have to like, you do have to switch out some things. Like, for instance, the MiniMax M3 head uses full attention, and with full attention you end up with this like insane bottleneck in spec dec ‘cause you're doing auto-regressive token generation for three tokens, and you're doing this like N squared over all of the tokens that are in your sequence. Your KV cache is like very large because it's not sparse, it's not top K. So we find it better to like, okay, we're gonna replace this, we're gonna replace this layer with a layer from another model that's using like GQA, for instance. And then just with the right training, you can get it to have the same acceptance rate. So it is very possible to retrofit layers from other models and very much needed. If a layer is like inefficient, the training just becomes the challenge, like how do you ensure that you train it properly? Which again to your earlier point is like the mesh between training and inference. As in like you need very good training in order to do fast inference. That's like, I feel like more and more becoming true.Swyx [00:20:21]: Yeah. Anything else on the support side when you say like get it to fully production ready?Loop Detection, Race Conditions, and Non-DeterminismPhilip [00:20:26]: Yeah. I think that there's also a question of just, we can test a model to a pretty extensive degree, but we're trying to get it out quickly and then you see a bunch of other people test it and you get interesting results. There was an issue with, GLM briefly where we had some like mode collapses where it would just output the same token over and over again for certain prompts on certain temperatures. Like once you expose an endpoint to the real world, there's going to be, so many more varieties of things given to it that you're able to, discover and patch things. So it's not just a, day zero process, it's then like for the first week, for the first month, if a model remains popular, like how do you both fix bugs and then continue to push the envelope on performance?Ali [00:21:21]: What do you mean you don't want your model outputting S?Swyx [00:21:24]: Is there loop detection on that stuff, by the way? It still happens like quite a lot, which is surprising.Ali [00:21:30]: We have like we, in our endpoint, like if a model was to output the same token like four plus times, we just cut the generation. We say like, “Oh, sorry, this-- Like try again,” or like we will reprocess the request. ‘Cause we know then, like if it, like if, yeah, it's four times the same token, it's probably collapsed.Swyx [00:21:45]: Yeah. Is there a way to opt out in case I really want that?Ali [00:21:48]: You want that?Ali [00:21:50]: I think there's a way that we have to handle it. I'm not exactly certain, but I feel like in certain models, like when they output something like you can imagine, like a table for instance, and so they want, they wanna draw like 12 dashes and 12 dashes. Yeah, I think there's a way for that to happen. I think we only do it on certain tokens. Like we exclude certain special characters.Swyx [00:22:07]: Yeah.Ali [00:22:07]: So we only do it on like certain like S is the most common almost. GLM-5.2Swyx [00:22:11]: OhAli [00:22:11]: And I think it was DSV 4 as well. Like you'd just have like looping issues where like you literallySwyx [00:22:17]: ItAli [00:22:17]: Just have like S.Swyx [00:22:18]: Yeah. Is there a special, something special about S? No, just randomlyAli [00:22:21]: It just seems to be the one token involved.Swyx [00:22:23]: Yeah. And it'Philip [00:22:24]: Is thereSwyx [00:22:24]: And it's only temperature 0Ali [00:22:27]: NoSwyx [00:22:27]: Even at other temperaturesAli [00:22:27]: Even at like 0.9 or whatever, it will still, it will still collapse.Swyx [00:22:30]: That's weird, right?Ali [00:22:30]: It's, it is an inference problem to be honest, like a software problem. Like oftentimes, the image you run will-- like NVIDIA will release an image for instance, and if we will upstream the changes from their latest TensorRT-LLM image into our stack, we'll find that it fixes it. Or oftentimes this will only happen in an inference engine that you're using like SGLang. But if you were to switch to vLLM, that isn't the case. So it seems to be like an extremely like deterministic software issue and not really a model issue. It's not like a weights problem. Like I'- we'll say like, “Oh, it's a problem with the quant. We did PTQ wrong,” right? But that isn't, that doesn't make sense because the same weights used with a different inference engine does not repeat the problem. And sometimes it's, the kernels that are being used in the backend have like these very subtle sometimes race conditions, where if you were to use this model hosted on one cluster, you will never get this problem.Swyx [00:23:19]: Oh my God.Ali [00:23:19]: But if you host it on a different cluster, you will. And the reason is the KV cache transfer from a node to node in that one cluster is using a slower interconnect than the node to node in another cluster. So that exposes the race, whereas in another cluster it doesn't. So then you end up just like, okay, this model is not gonna be hosted on this cluster. We're gonna host it on, another cluster because that cluster exposed that problem. But then it ends up with like, okay, is it the software? Is it the model weights or is it the hardware?Swyx [00:23:42]: There is a thing about this with temperature 0 still not being deterministic, right?Ali [00:23:46]: Right.Swyx [00:23:46]: Mostly because of hardware. Even at temperature 0 same model, you won't always get the same output.Swyx [00:23:52]: Even-- But I'm surprised by the race condition one because, I thought PyTorch was a graph that like guarantees that you at least, execute things in the right order.Ali [00:24:02]: Well, yeah, true. Like I'm not, I'm not saying that there is. Like well, you have things like PTL optimizations where like you can start a kernel before the end of the previous kernel, and that's like ‘cause you want to do that because there'sSwyx [00:24:12]: It's like pipeliningAli [00:24:12]: Expense. Exactly.Swyx [00:24:13]: Yeah.Ali [00:24:13]: But it'- But you don't do it cleanly. Like you overlap a little bit of the execution. No, it is very possible that the kernel itself, like that one block that is supposed to be running in this instance of time, that kernel itself has a race condition. For instance, like a missing barrier. Like often if you're designing a kernel and you want it to make it to be very fast, if you don't test it extensively, you'll, you'll have certain threads access data points from registers before they've been written to by other threadsSwyx [00:24:36]: YeahAli [00:24:36]: For example, because like your barrier is wrong or your synchronization was wrong. But yeah, like the testing itself is very difficult in those like, andSwyx [00:24:42]: And there's no like borrow checkerAli [00:24:45]: What does that mean?Swyx [00:24:46]: Like Rust. Like the. If you're trying to have like memory safety It sounds like a comparable problem.Ali [00:24:52]: Well, yes, but you're working in CUDA, right, NVIDIA GPUs. Like- You just need a higher level language like modular Maybe that's what modular is supposed to do. I don't know.Quantization Quality and Vendor FidelityVibhu [00:25:00]: How do you see keeping quality of the model? So you talked about all these steps of, okay, you gotta do quantization, train your own speculative decoderAli [00:25:07]: RightVibhu [00:25:07]: Run on different hardware. Looking at other model providers, okay, you kicked off a inference speed race on the consumer end. What goes into keeping quality the same across them, right? Sure, you can run benchmarksAli [00:25:22]: YeahVibhu [00:25:22]: But, like, how do you determine how much quantization are there standards? What goes intoPhilip [00:25:27]: There's a few things on quality. Most inference optimizations are lossless. KV caching, for example. You are just recomputing or preventing recomputing the same values. Speculation, of course, if a draft token is wrong, it gets rejected. The main lossy optimization is quantization. And that really comes down to, number one, data format, number two, which parts of the model you choose to quantize, which layers, and number three, like doing a lot of calibration on the quantized weights, to ensure that you're preserving all the outliers. There's other tricks that you can do, though. A big one is long context, ‘cause one thing you asked at, right at the beginning is, “Oh, what's gonna happen if I send a 200,000 token request in?” So with a long input sequence, you need to, store a lot more information. You need to process a lot more tokens. And so even if a model has a context of a certain length, you might, as an inference provider, choose to build an API with a shorter context length, and of course a full length one as well. Because if someone doesn't need the full million token context, for example, you can get them better performance. I don't know if that's exactly like quality of the model. The way that I think about quality is to what degree are we faithfully serving the original model? If you think of a golden implementation of a model that performs exactly the way the model is designed to perform, I think of quality as how close are we getting to that, 100% fidelity of the model.Philip [00:27:13]: You can also, of course, think about quality from the training side and how do you push yourself past 100%. But when I think about purely inference optimizations, it's getting faster while staying as close to that 100% fidelity mark as possible. And certainly our standard internally is that, like you should not be able to tell the difference between our API and a, official API. I think Kimi in particular does a good job of vendor benchmarking hereAli [00:27:41]: YesPhilip [00:27:41]: Where they haveAli [00:27:42]: They released an actual vendor benchmark.Philip [00:27:43]: Exactly, yeah.Ali [00:27:44]: ‘Cause they accused, some people, Amazon? There was some provider that was not doing very well on Kimi's benchmark.Philip [00:27:50]: Yeah.Philip [00:27:51]: So, with Reflect we probablyVibhu [00:27:52]: This was a long time ago, right?Philip [00:27:54]: No.Ali [00:27:54]: Yeah, like threeVibhu [00:27:55]: They alsoAli [00:27:55]: Four, five months agoVibhu [00:27:57]: This also happened with, I don't remember which model, but they pulled out quite a few, and then they started a whole chart about this. It might have beenPhilip [00:28:03]: Kimi Vendor Verifier.Ali [00:28:04]: Yeah.Philip [00:28:05]: Yeah.Ali [00:28:05]: Yeah, ‘cause you, ‘cause you'd be pissed, right? Like if you'Philip [00:28:07]: Yeah.Ali [00:28:07]: If like if I'm a consumer and I'm using like Amazon's endpoint for instance, and I've used Kimi and I'm like, “Oh my God, like this is bad,” I'm not gonna say, “Oh, Amazon quantized the model in a bad way.” I'm gonna say, “Oh, Kimi sucks.” Right?Philip [00:28:17]: Yeah.Ali [00:28:17]: So it seems like that makes sense.Philip [00:28:19]: Yeah, they care. They care.Vibhu [00:28:21]: Justifiably.Ali [00:28:21]: Yeah, justifiably.Vibhu [00:28:22]: This is probably a stupid question, but just checking, has anything improved from main quantization?Philip [00:28:28]: Yeah.Vibhu [00:28:28]: Like, is quantization always strictly worse?Ali [00:28:30]: Well technicallyVibhu [00:28:32]: NoAli [00:28:32]: It's a lossy. QuantizationPhilip [00:28:33]: YeahAli [00:28:33]: Is a lossy, it's a lossy implementation.Philip [00:28:36]: Speed improvesVibhu [00:28:36]: Speed improves.Ali [00:28:37]: It the number, likeVibhu [00:28:38]: No, I' always look for inverse scaling laws.Philip [00:28:40]: Yeah.Ali [00:28:40]: Yeah.Vibhu [00:28:40]: This is something I learned from Noam Brown, where like things that normally act in one direction sometimes do.Philip [00:28:45]: Well, technically when you run a benchmark, because these models are deterministic, sometimes your,Ali [00:28:52]: YeahPhilip [00:28:52]: NVFP4 quant is like, two basis points higher than yourAli [00:28:56]: No, it's noise. It's noise.Philip [00:28:57]: Yeah, exactly. I'm like, yeah, it's, it's within. That's why I always say within margin of error.Philip [00:29:01]: And I stopped saying that because everyone assumes that what is, well, within some margin of error, we're barely inside of that to the worst, so we're saying. But yeah, sometimes it's just like, gives you a higher output score. But like Ali said, that's noise. To my knowledge, you're not necessarily making the results better. You're just trying to, again, like keep your fidelity as close to 100% to the original model.Layer Selection, KL Divergence, and Better QuantizationAli [00:29:27]: There is, to your point, research that we did on MP. I don't know if you are able to pullPhilip [00:29:31]: YeahAli [00:29:32]: A tweet we did. One of our research interns, Joshua, I think it's a tweet on how we have 20% better quantized GLM-5.2 than NVIDIA. Essentially what we found throughout like this month research is, okay, quantization is a lossy. It's. You're compressing the data from, occupying 16 bits to occupying, four bits, for instance. And so you're losing some information, and you're trying to minimize that. And so when I say that I'm gonna quantize the model, my job becomes how do I find the layers that I can quantize, and how to find the layers to not. For instance, with image models, I don't quantize modulation layers, and I don't quantize out projections because those two are. Like out projection is what you see as the user. Modulation is what the model sees or understands. Right, exactly. And so to his paper, do you have the. It doesn't have the. Yeah. It's a long paper. I don't know if I can findVibhu [00:30:25]: If there's a part to search or it's probably in the thread.Ali [00:30:28]: It's probably in the thread.Vibhu [00:30:29]: Yeah.Ali [00:30:29]: But the long and the short is it is very possible that quantizing more of the model makes the results. Like if I have a model that I quantize layers one, five, and 10, and another model where I only quantize layers one and It is possible that the model in which I quantized more information is going to perform better because the quantization errors have canceled out. And so what Joshua showed in his mathematical proof where he had like a verifier in, is that you can predict which layers are going to have quantization errors that will cancel out with each other, and you choose to quantize those layers. And so the result of doing this mathematical quantization is you end up with a model that's 20% more quantized than another provider, so you get 20% more throughput of it because there's more layers than running an NVFP4, and your quality is better than that other quant because the layers that you chose to quantize have their errors cancel out, like one layer skewed to the right one layer skewed to the left, one layer skewed to the right. Your final logits distribution is more similar to the original distribution of the model, so you have better fidelity. And so the way we proved this was with KL divergence. So instead of just scoring on the benchmarks, we scored the KL divergence between the logit distribution of the quantized model and the logit distribution of the original full precision model, and we showed that with this technique we get. If your probability distribution on the logits which token it wants to select is more of the same as the original model, you're probably gonna end up staying true to the original model. So yeah, so it seems like previously before this, it seemed like the industry was, well, the more you quantize, the worse it's gonna be, ‘cause the more loss you introduce. That's not exactly, not necessarily true. So yeah, doesn't improve it, but can cancel out.Philip [00:31:57]: I think it might be this, but reminds me a good bit about pruning where you can prune off certain layers.Philip [00:32:03]: But very interesting. Didn't know this was a whole paper you guys put out.Ali [00:32:06]: It's. Fun fact, it was originally 72 pages, this paper, and then we decidedPhilip [00:32:11]: WowAli [00:32:11]: We can't tell. We couldn't release it. So it's now 45.Swyx [00:32:15]: Still 39 pages, so very substantive. We talked about evals and all these things and, like what's possible in terms of speedup? Like it's like probably like the numberInference Speedups and BenchmarkingSwyx [00:32:25]: Thing that people do wanna care about, and it's something that you wrote about in your post. Like official API is 70 tokens per second, and you push it up to 90. Is that like a normal thing?Philip [00:32:36]: So what's cool about working in inference, the reason that I think inference is going to be a useful place to do engineering for a long time, is that if you look at highly optimized domains like, say, finance, if you're in finance, you measure how much better you got in basis points. It's like, “Oh, I got five basis points better, like twentieth of 1% better,” that's huge news because everything is so optimized. When we publish optimizations, it's 20%, it's 100% it's 200%. So there's still probably like a lot further to go, honestly. Like you'll, you'll know that inference is pretty much solved when researchers start publishing about how they got 1% faster at something.Swyx [00:33:19]: Which by the way, because I am from the finance background, in the ‘70s, that was the margin at the time. When you did quantitative finance research, you would findAli [00:33:27]: And like 20%, tens of percent.Swyx [00:33:29]: That's. Yes.Philip [00:33:29]: Yeah.Swyx [00:33:30]: And now it'Philip [00:33:31]: Tiny fractionsSwyx [00:33:32]: For those people interested, look up Andrew Lo's paper. He had a really interesting illustration of quant, stat arb, distribution, narrowing down from like those kinds of 20% differences in the ‘70s, down to nothing today, which is very cool.Philip [00:33:48]: Exactly, and we're at the beginning of the same type of thing. Now benchmarking is hard. I think anyone will tell you that, and benchmarking provider speeds is hard because there's so many variables that go into it. What hardware are you using? How much load do you have on the system? What's the exact nature of the prompts and input and output sequence lengths? All that stuff. But overall, when you start stacking these improvements, you're looking at multiples. You can look at it. The most common form, of course, is TPS, tokens per second, which is bad naming by us in the industry, ‘cause there's two tokens per second. There's tokens per second, the throughput number, and the latency number.Ali [00:34:31]: TTMT, yeah.Philip [00:34:32]: Like total tokens per second out of the, out of the GPU as a throughput number. Most people only care about tokens per second as the latency number, which we should call ITL, intertoken latency, but we don't.Philip [00:34:44]: Anyway, so you can imagine a standard API without many optimizations for a 1 trillion parameter model operating somewhere in the 30 to 50 tokens per second range for reasonable traffic profile. And we generally see the goal of, pushing to 10X that. But, not necessarily day zero, but by stacking enough optimizations, if you have, say like four optimizations, each of which doubles performance. Or sorry, three optimizations, each of which doubles performance, then you stack that up, that's an 8X gain. That's the order of magnitude that we're working with in this space. We're trying to make things substantially faster, not just go from like 70 to 90.Swyx [00:35:38]: Are you saying you've. You have done that?Philip [00:35:40]: So let's say you have as a reasonable baseline, 30 or 40 tokens per second. You can achieve 10X that. So like on GLM-5.2, if you run it unquantized, perhaps on H100s even, and you're just using an off-the-shelf inference engine with no particular optimizations, no speculator, nothing extra around like KV routing, no disaggregation, you're, you're probably, yeah, looking at that like 30 to 40. You think that's like a reasonable baseline?Swyx [00:36:12]: Right. Right.Philip [00:36:12]: To get to something like 10X, there's a lot of trade-offs that you're making. If we're running at more like a 300, 400 tokens per second range, you are using the best hardware possible. You have a optimized speculator. You have done all of your quantization work. You are Seeing a pretty high cache hit rate. You are running with a reasonably small batch size and a parallelism configuration that is tuned for latency versus throughput, but it is possible. So the spreads that you see if you, like, go on artificial analysis or you go on OpenRouter and you look at, the worst provider to the best provider, oftentimes can hit that range. 10X is of course very aggressive. It's oftentimes maybe more of a four to six times improvement. But that's the performance that makes us really excited, is when we can get these huge gains, not just go from 70 to 90 tokens.Stacking Optimizations: NVFP4, Speculation, and DisaggregationAli [00:37:19]: It's also, like, hardware dependent. Like, ifPhilip [00:37:20]: YeahAli [00:37:20]: If you have a thing where you're serving it on just, like, a node of H100s and then you throw, like, you shard the model across, like, four nodes of B200s. Like, you can definitely increase the speed with just throwing more hardware at it. Like, normalizing for the same exact hardware and the same number of GPUs.Philip [00:37:35]: Yeah. Then you're looking at, like, a two to 4X improvementAli [00:37:38]: Right. RightPhilip [00:37:38]: Depending on the inference optimizations. So yeah, it's. Some of it's, what's the call, and some of it's who's the driver.Vibhu [00:37:46]: If you break down the two to 4X, say the example is run GLM-5.2Ali [00:37:51]: YeahVibhu [00:37:51]: On B200sAli [00:37:53]: YeahVibhu [00:37:53]: Single node, right? What's, like, the cost trade-off for effort to get, like, the last bit of juice out versus what should people just think of, right?Ali [00:38:01]: Spectre quantization. Yeah.Vibhu [00:38:03]: Spectre quantization.Ali [00:38:04]: That's, that's, that's like 95%. LikeVibhu [00:38:06]: And how far does that get you? And how easy is that for the average person to do? So say right I wanna throw the weights of GLM-5.2 on a node of B200s, how easy is it to find speculative decoder- decoder model or already quantized model? How much work goes into it?Philip [00:38:23]: If you're doing it up front, it's quite a lot of work. If you're doing it today, there's going to be people who have published things that you can just, you can just grab some NVFP4 weights. You can grab a speculator. Yeah, if we're thinking about, like, what are the 2Xs we're stacking, going from, BF16 to NVFP4 is, it's not quite a 2X, right? It's like. I think it's about, like, 30 to 40%, from 16 to 8, and then another 30 to 40% multiplied from, 8 to 4. So that doesn't quite get you a 2X, but, like, roughly a 2X. Speculator, roughly a 2X. Disagg on top of that if you're able to get enough hardware and put enough traffic through it, another roughly a 2X. And then you add in some, double-digit percent increase from having just a better runtime with, the latest kernels and stuff behind it. And that's how it stacks up.Ali [00:39:21]: YeahPhilip [00:39:21]: So building each of those, like, building the, quantized weights is, for someone who really knows what they're doing, hours to days of work. Building the speculator, again, like, hours to days of work. And the, disagg setup, hours to days. Well okay, but like once you haveAli [00:39:39]: Once set up. Once set up. YeahPhilip [00:39:40]: Yeah, getting disagg working for the first time, I'm saying, of course, is very difficult.Philip [00:39:44]: The marginal implementationAli [00:39:48]: Like, if you're just grabbing, like if you are a person, like just a normal consumer who has access to, like, a node of B200s and you're wondering, “How can I just host it myself?” You don't need to quantize the model yourself. There's always gonna be, like, an open source quantized checkpoint. NVIDIA's gonna push one out if no one else does. You. Usually, the providers will have their own spec dec that they've trained as well. You don't need to train your own spec dec. You can just use that as well.Philip [00:40:09]: Yeah. Like, GLM-5.2 has its own MTP.Ali [00:40:13]: Right. Right.Vibhu [00:40:14]: What's multi token prediction?Philip [00:40:15]: Yes.Ali [00:40:16]: I'm justVibhu [00:40:16]: Can you explain that?Ali [00:40:16]: I'm just an expert.Ali [00:40:18]: I can do it for you in case I get it wrong?Vibhu [00:40:20]: No.Vibhu [00:40:21]: Yeah, you should correct if we're wrong, but their multi-token prediction can be used for self-speculative decoding.Ali [00:40:27]: I'm not sure. I'm not gonna correct that.Vibhu [00:40:28]: Okay. I'm semi-confident in thatAli [00:40:30]: Okay. YeahVibhu [00:40:30]: But someone can check. But it's useful to paint the story of, okay, not just the average person, but say a company wants to switch from serverless inference I wanna throw this up on. I wanna rent some GPUs, throw it up. These are the steps you take to do significantly faster than just put it behind vLLM.Ali [00:40:48]: Right.Vibhu [00:40:49]: I was waiting for a mention of Dynamo.Vibhu [00:40:51]: I feel like, that's supposed to be the baseline that you measure against.Dynamo, KV Routing, and Disaggregation ToolkitsPhilip [00:40:55]: I would think of Dynamo as less of a box system and more of a toolkit for building with. So when we talk about doing aware routing, when we talk about doing KV offloading, when we talk about doing, PD disaggregation, Dynamo fundamentally is. By the way, Dynamo is an open source library from NVIDIA.Ali [00:41:17]: We've done a pod with KylePhilip [00:41:18]: OkayAli [00:41:19]: Kyle Cranin.Philip [00:41:19]: Cool. So then your listeners know then that it supports all the different inference frameworks. And it is multi hardware, which is interesting.Ali [00:41:28]: But it's just a router, it's not like an optimizer layer.Philip [00:41:30]: Yeah. All it does, like, what Dynamo is good at, it is a library for moving information around your cluster, around your hardware. So if you have, KV cache on one place and you need it to be somewhere else, Dynamo coordinates NIXL for you to move that around.Philip [00:41:49]: That doesn't mean that, like, out of the box, you just say, “Pip install Dynamo,” and then you get, like, a massive performance speed up. It's more of a developer toolkit.Ali [00:42:01]: Yeah. I would have said it would. It comes with a set of defaults that you can then swap out.Philip [00:42:06]: It does. If the industry at large, I think, was, like, rolling out all of these deployments, standard, then I think it would be, like, a credible baseline. But, we've got to, we've got to benchmark against, like, what we're seeing in the wild.Speculative Decoding Methods: Medusa, EAGLE, n-Gram, and Spec-SpecVibhu [00:42:23]: I did wanna talk a little bit more about PD disagg, because that is probably, like, number three after quantized and speculative decoding. In your book though, I was just gonna pull out the book.Philip [00:42:31]: Yeah.Vibhu [00:42:32]: Like section 522 on Medusa, 523 on EAGLEPhilip [00:42:35]: YeahVibhu [00:42:36]: 524 on gram.Philip [00:42:37]: It's 55, would be disaggregationAli [00:42:42]: Yeah. Well, no, I just wanted to dwell a little bitPhilip [00:42:44]: YeahAli [00:42:44]: The other. Like, so what do you choose to include? What do you choose to not to include? Because there was all these other techniques.Philip [00:42:51]: Yeah.Ali [00:42:51]: Are these still relevant? Because I think they came out, like, a year and a half ago maybe.Vibhu [00:42:55]: Medusa is quite old.Philip [00:42:56]: Yeah, Medusa's old.Ali [00:42:58]: It was old.Vibhu [00:42:58]: But is it in the book as a good, here'sPhilip [00:43:01]: BaselineVibhu [00:43:01]: Baseline vanilla understand it?Philip [00:43:02]: Like you should know this.Vibhu [00:43:03]: Like I read the paper, I'm like, “ it makes so much sense.”Philip [00:43:05]: Yeah.Philip [00:43:05]: So with the book, I had a couple goals. One was to give people just a working vocabulary for the space as a whole, and the other was to give them some intuition about how each of these techniques works. As I mentioned in my AI Engineer talk, which is the first public addendum to this, the speculation space has moved much faster than everything else. So yeah, even at the time that I wrote the book Medusa, I very much included as a way for people to understand how the space evolved rather than what the most modern technique is. And now of course, there's DFlash, dSpark. There's, there's newer techniques even than EAGLE, although EAGLE is still very commonly used.Ali [00:43:51]: SpecSpecta.Philip [00:43:52]: Yes. Speculative decoding.Vibhu [00:43:54]: What canAli [00:43:56]: Oh, it's a paper by Tri Dao and it's like, it's doing speculative decodingVibhu [00:44:00]: HuhAli [00:44:01]: For the speculative decoder.Philip [00:44:02]: Oh, in spec- oh my God.Ali [00:44:02]: It's literally just an another. It's like, yeah, that's the most simple way to explain it, and it seems like he got trivial speed ups there. But it seems that the complexity with training, it's almost like in our mind at least, it's almost as complex as training GANs. Like it's like a very delicate balance and oftentimes you, it's just but yeah, it's literally speculative decoding on speculative decoding.Vibhu [00:44:21]: Speculative.Ali [00:44:22]: Yeah. We saw this paper.Vibhu [00:44:24]: It's interesting, right?Ali [00:44:24]: Yeah.Vibhu [00:44:24]: I wouldn't even expect it to be very particular to train, I wouldAli [00:44:29]: Right.Vibhu [00:44:29]: The naive part of me is like, okay, train speculative decoder.Ali [00:44:32]: But like, and it makes sense, like the whole idea of speculative decoding is you. It's like, it's like almost like the iPhone auto predict version but for a normal model, right? Like you're just, you're just, generating three tokens and you're like, okay, I'll do prefill on them. And so you save those three turns for your original model. Now your speculative decoder is doing three turns of auto regression, so why not just have an even smaller model?Ali [00:44:53]: The other question there is what are the size of speculators? So say forPhilip [00:44:58]: Right. It's like a billion parameters.Ali [00:45:01]: Like for MiniMax, it's. Yeah. It's like one layer. It's like one 60th of the original model usually.Philip [00:45:06]: Yeah. I think we should do a paper when we get back to the office.Philip [00:45:10]: SpeculativeAli [00:45:11]: SpeculativePhilip [00:45:11]: Decoding.Ali [00:45:13]: No, it's, it does seem like how, when do you stop? But then it also seems like if you're able to train spec-spec decode for instance, right? Like if you're able to have a small model that is accurately predicts what the intermediate speculator is gonna predict, that is able to predict what the original target model's gonna predict, then why not just use that smallest model directly, right?Vibhu [00:45:34]: Yeah. This isAli [00:45:35]: Like it seems likeVibhu [00:45:35]: Adjacent to the routing problem.Ali [00:45:36]: Right.Vibhu [00:45:36]: Yeah.Ali [00:45:36]: Right.Philip [00:45:37]: The thing with speculators is one of the practical constraints on using them is that you do have to run a small model on the same hardware that you're running the big model on. There is a orchestration and resource competition problem inherent in that, and that is one of the constraints on speculation in general, is that draft tokens cost resources to create and cost software complexity to manage. And so if you have like infinitely recursive speculators, you add in quite a bit of that complexity on the actual implementation within the inference engine as well, not just in the training process.Vibhu [00:46:17]: I was gonna say, I would wonder if you could do similar, like distillation and pruning of, it's the same thing, it's just a model. Can we not just distill a lot of the weights, quantize the speculator, out of my domain? The question that also comes up is, this is all for big server workloads, right? How much of this applies to, say I have this MacBook, I wanna run Gemma really efficiently. Similar problems, not the same?Local AI vs. Data Center InferencePhilip [00:46:45]: Pretty different. I talked to Selo, about this on his podcast a couple weeks ago. The difference between inference engineering for the data center and for production workloads versus inference engineering for local AI, is that we start with fundamentally like different constraints and different goals. With local AI, it's how do I fit this model onto my hardware and then make it less dumb? And with data center influence, it's how do I load this model and then make it less slow? And we care about less dumb, and they care about less slow. But the local AI inference engineering ecosystem, I think has a lot for us to learn from in the data center space. They are experts in various forms of quantization, including dynamic quantization that we just don't touch, in the pruning, in the distillation, in the, layer removal. There'Ali [00:47:42]: Layer removal matters less.Philip [00:47:43]: Yeah. There'Ali [00:47:44]: No one loves pruning really.Philip [00:47:45]: Yeah. Well, but the, but they doVibhu [00:47:46]: Which is surprising, right? But that's, that's a whole different thingPhilip [00:47:48]: Just to fit something on the laptop.Ali [00:47:50]: Right.Philip [00:47:50]: So yeah, it's a, it's an interesting, it's an interesting space. Not necessarily that like their techniques make sense for us to do in the data center, because we have different resources and different goals, but more that the process as well as the openness of that field is something to, admire.Ali [00:48:12]: Yeah. Like to your point, like, certain optimizations that would. Like for instance, Turbo Quantum Sharper, like it made such huge hype on that and we did like a whole deep dive on Twitter and like said, what is it? How does it work? Why is it good or not? And it took off and it was implemented on local devices because your memory bandwidth is so slow on like a MacBook, for instance. But try putting the same thing on like an NVIDIA GPU on a B200 Turbo quant would not be. Like, it would not be used. Like, NVIDIA - Like, NVIDIA made it clear that this is not a good optimization, and we've seen it firsthand where the overhead of doing dequantization, quantization of, in the kernel itself with turbo quant kernel, each end is much slower than the time that you save from doing the bandwidth. ‘Cause on the B200s, you have like 3.5 terabytes per second. You don't need decrease the storage that much. You don't need to do, FP4 KV cache. You don't need to use a requant. There's, there's, there's better optimizations to be made. But on Edge devices, it's extremely important, it's extremely useful. So, seems to be, like, different optimizations there, but then they're all uniquely combined with like all you wanna quantize the model, you wanna do speculative decoding, like certain common prefixes with bothPhilip [00:49:18]: Principles.Ali [00:49:19]: Yeah, exactly. Exactly. Exactly.Philip [00:49:20]: They also do a lot of work on, model parallelism, especially over, heterogeneous topology, where you have, some sparks and they are wired together with, Ethernet, DGX sparks.Ali [00:49:35]: Yeah, this is the Exo Labs guys.Philip [00:49:36]: Yeah. You have, a nu

Matt Modi is joined by Andrew Erickson and Scott Bogman to break down their favorite fantasy football WR targets across every stage of your draft. From early-round stars like Zay Flowers and Jaylen Waddle to mid-round breakout candidates including Parker Washington and Marvin Harrison Jr., plus late-round sleepers like Trey Harris, Jayden Higgins and Antonio Williams, the crew identifies the receivers who could dramatically outperform their ADP this season. Whether you're searching for WR1 upside, undervalued draft values or deep sleepers that can win your fantasy league, this episode has you covered. Timestamps: (May be off due to ads) Intro - 0:00:00 Zay Flowers - 0:02:03 Mike Evans - 0:05:33 Jaylen Waddle - 0:08:14 Carnell Tate - 0:12:42 FantasyPros Premium Giveaways - 0:14:23 Marvin Harrison Jr. - 0:15:15 Parker Washington - 0:22:08 Chris Godwin - 0:27:41 Jordan Tyson - 0:31:30 FantasyPros Draft Wizard - 0:34:15 Wan’Dale Robinson - 0:35:05 Jayden Higgins - 0:40:03 Tre Harris - 0:44:32 Antonio Williams - 0:53:08 Hard Rock Bet - 0:58:25 Outro - 1:00:25 Helpful Links: Hard Rock Bet - Sign up for Hard Rock Bet and make a $5 bet and you'll get $150 in bonus bets if you win. Head over to Hard Rock Bet, sign up and make your first deposit today. Payable in bonus bet(s). Not a cash offer. Offered by the Seminole Tribe of Florida in FL. Offered by Seminole Hard Rock Digital, LLC, in all other states. Must be 21+ and physically present in AZ, CO, FL, IL, IN, NJ, OH, TN or VA to play. Terms and conditions apply. Concerned about gambling? In FL, call 1-888-ADMIT-IT. In IN, if you or someone you know has a gambling problem and wants help, call 1-800-9-WITH-IT. GAMBLING PROBLEM? CALL 1-800-GAMBLER (AZ, CO, IL, NJ, OH, TN, VA) Draft Wizard - Dominate your fantasy football draft with Draft Wizard. Run fast mock drafts, test different strategies, build custom cheat sheets, get pick-by-pick draft advice, and learn your leaguemates' tendencies before draft day. Just download the FantasyPros App or head to fantasypros.com/draftwizard Follow us on Twitch - The team here at FantasyPros is taking questions all week, every week on Twitch. Follow us on Twitch at twitch.tv/fantasypros and never miss a stream! Discord – Join our FantasyPros Discord Community! Chat with other fans and get access to exclusive AMAs that wind up on our podcast feed. Come get your questions answered and BE ON THE SHOW at fantasypros.com/chat Leave a Review – If you enjoy our show and find our insight to be valuable, we’d love to hear from you! Your reviews fuel our passion and help us tailor content specifically for YOU. Head to Apple Podcasts, Spotify, or wherever else you get your podcasts and leave an honest review. Let’s make this show the ultimate destination for fantasy football enthusiasts like us. Thank you for watching and for showing your support – https://fantasypros.com/review/ BettingPros Podcast – For advice on the best picks and props across both the NFL and college football each and every week, check out the BettingPros Podcast at bettingpros.com/podcast, our BettingPros YouTube channel at youtube.com/bettingpros, or wherever you listen to podcasts.See omnystudio.com/listener for privacy information.

As more applications and services rely on AI inferencing, that means more traffic on the WAN. Yes, bandwidth can solve a lot of problems, it's not unlimited and AI inference traffic still has to share pipes with other flows. That means things like WAN latency, determinism, security, and traffic engineering are just as important as... Read more »

As more applications and services rely on AI inferencing, that means more traffic on the WAN. Yes, bandwidth can solve a lot of problems, it's not unlimited and AI inference traffic still has to share pipes with other flows. That means things like WAN latency, determinism, security, and traffic engineering are just as important as... Read more »

As more applications and services rely on AI inferencing, that means more traffic on the WAN. Yes, bandwidth can solve a lot of problems, it's not unlimited and AI inference traffic still has to share pipes with other flows. That means things like WAN latency, determinism, security, and traffic engineering are just as important as... Read more »

professorjrod@gmail.comIf “the internet is down” is the sentence that makes your brain freeze, this guide is for you. We take the pressure off by turning networking into a set of simple ideas you can picture and repeat under stress, the same way you'll need to think on the help desk and on the CompTIA A+ exam. I walk through what a network is, why businesses rely on it, and how PAN, LAN, and WAN show up in real life, from Bluetooth devices to the internet itself.From there, we connect the dots across the gear and the settings: what switches do inside a LAN, why routers matter for getting online, how modems talk to your ISP, and what an access point really does for Wi‑Fi. We cover IP addressing with clean examples, the difference between private and public IPs, and how NAT translates between them. Then we hit the “must know” services: DHCP for automatic configuration, DNS for turning names into numbers, and the core ports that show up again and again. I also share a quick help desk story that proves why “never guess, always verify” saves time.The second half expands beyond networking into the wider set of IT support fundamentals: laptop design and upgrade limits, battery safety, modern connectors like USB‑C power delivery, docking stations, mobile tech like NFC and biometrics, and printer troubleshooting that becomes easy once you understand the printer type and the laser process. We wrap with practical explanations of virtualization and cloud computing, including IaaS, PaaS, and SaaS, plus why the smartest solution is often the right service, not more hardware.Subscribe for more CompTIA A+ and IT certification study help, share this with a friend who's studying, and leave a review so more new techs can find the show.Support the showArt By Sarah/DesmondMusic by Joakim KarudLittle chacha ProductionsJuan Rodriguez can be reached atTikTok @ProfessorJrodProfessorJRod@gmail.com@Prof_JRodInstagram ProfessorJRod

Optics is no longer a supporting accessory in the data center network. As AI infrastructure advances from 400G and 800G toward 1.6-terabit connectivity, optical components are consuming a larger share of network cost, power and operational risk. In this episode of the Data Center Frontier Show, DCF Editor in Chief Matt Vincent speaks with Bill Gartner, Senior Vice President and General Manager of Cisco's Optical Systems and Optics business, about how AI is changing the strategic role of optics. Gartner explains that optics represented roughly 10% of a network port's bill of materials at 10G. At 400G and above, the optics can cost more than the switch port itself. Reliability has also become critical: A single unstable link can force GPUs operating in parallel to stop, return to a checkpoint and restart. According to data Cisco has seen from hyperscale customers, link flaps can reduce GPU infrastructure efficiency by as much as 40%. The conversation maps the AI network across three distinct tiers: Scale-up: Connections within the rack, carrying approximately 500 times the bandwidth of a traditional WAN environment. Scale-out: Connections between racks, commonly using 400G and 800G pluggable optics. Scale-across: Coherent optical connections between data centers as AI clusters expand beyond the power limits of a single facility. Gartner also discusses Cisco's 1.6T roadmap, routed optical networking, coherent pluggable optics and the emerging debate around co-packaged and near-packaged optics. These architectures promise lower power consumption and greater density, but introduce new questions involving interoperability, replacement and operational resilience. Looking ahead, Gartner emphasizes that optics is not constraining AI network growth. It is enabling clusters to scale across racks, campuses and geographically distributed data centers, while the coming inference wave shifts the industry's focus toward cost and power efficiency.

Check out the Shokz OpenDots 2 clip-on ear buds and use code 'WAN" to get $15 USD or $20 CAD off https://beopen.shokz.com/WANSHOW_Opendots2 Go to https://bit.ly/4y9VZ0i to enter the sweepstakes to win one of the two brand-new PowerColor Red Devil AMD Radeon™ RX 9070 XT GPU, and an AMD LED light sign! Visit https://www.squarespace.com/WAN and use offer code WAN for 10% off Level up your streaming and recording game with XSplit Broadcaster! Get 30% off your first purchase or subscription with code WANSHOW30 at https://lmg.gg/xsplitwan Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

Who is the best fantasy pick between Brian Thomas Jr. and Parker Washington? Is Alec Pierce one of the most slept on players in fantasy football?1010XL's Trent Osborne and Jaguars Today host Tony Smith break down the entire AFC South on The XL Fantasy Show!Timecodes:0:00 - Intro1:38 - Jacksonville Jaguars Fantasy Football Breakdown8:43 - Indianapolis Colts Fantasy Breakdown8:48 - Which Jonathan Taylor Will We Get in Fantasy Football?13:24 - Tyler Warren Fantasy Football Breakdown16:38 - Daniel Jones Fantasy Football Breakdown19:57 - Alec Pierce Fantasy Football Breakdown26:58 - Houston Texans Fantasy Breakdown 27:08 - C.J. Stroud Fantasy Football Breakdown32:44 - Nico Collins Fantasy Football Breakdown34:20 - David Montgomery Fantasy Football Breakdown39:06 - Tennessee Titans Fantasy Breakdown39:19 - Cam Ward Fantasy Football Breakdown43:42 - Carnell Tate Fantasy Football Breakdown47:26 - Wan'Dale Robinson Fantasy Football BreakdownSubscribe to our YouTube channel! https://www.youtube.com/user/1010XLVideo?sub_confirmation=1Follow us on social media!►Twitter: https://twitter.com/1010xl►Tik Tok: https://www.tiktok.com/@1010xl►Instagram: https://www.instagram.com/1010xljax►Facebook: https://www.facebook.com/1010xlCheck us out wherever you stream podcasts!►Apple: https://podcasts.apple.com/us/podcast/1010-xl-jax-sports-radio/id1011442917►Soundcloud: https://soundcloud.com/1010xl-92-5-fm-jax►Spotify: https://open.spotify.com/show/4QMbFucyBE1NnCHes9A7CVFind all of our shows here! https://1010xl.com/listen/#jaguars #jacksonville #jacksonvillejaguars #duuuval #jaguarsfootball #jacksonvillefootball #nfl #football #travishunter #trevorlawrence #fantasyfootball #2026fantasyfootball #fantasyfootball2026 #fantasydraft #startsit #fantasyfootballdraft #draftstrategy #fantasyfootballtips #fantasyfootballvalues #fantasyfootballrankings #fantasyfootballadvice #nflfantasy #fantasyfootballinjuries #nflinjuries #fantasyfootballnews #fantasysleepers #colts #indianapoliscolts #titans #tennesseetitans #texans #houstontexans #afcsouth #cjstroud #jonathantaylor #nicocollins #camward #carnelltate

Chris Welsh, Andrew Erickson, Derek Brown, and Jake Ciely go head-to-head in a full PPR fantasy football mock draft using Draft Wizard. The crew breaks down how to attack a 2-WR format, why running backs rise in value, when to draft elite tight ends, and how to balance early-round stars with mid-round upside. Find out how the guys build around players like Puka Nacua, Christian McCaffrey, James Cook, Justin Jefferson, A.J. Brown, Chase Brown, Trey McBride, Colston Loveland, Jaylen Waddle, Garrett Wilson, Breece Hall, Christian Watson, Quinshon Judkins, D'Andre Swift, Trevor Lawrence, Kyle Pitts, Brock Purdy, and more. Plus, the Draft Wizard grades every team, reveals the best picks and biggest reaches, and shows how expert mock draft strategy can help you prepare for your fantasy football draft. Timestamps (May be off due to timestamps):Intro - 0:00:00Draft Order Banter & Pick Spots - 0:00:452-WR League Strategy - 0:01:56Why Running Backs Rise in 2-WR Formats - 0:02:24Use the Draft Wizard at fantasypros.com/draftwizard - 0:04:10Draft Slots Revealed - 0:05:06Round 1 Begins - 0:05:57Puka Nacua at 1.03 - 0:06:34Christian McCaffrey at 1.06 - 0:07:04James Cook at 1.11 - 0:07:48Justin Jefferson & De’Von Achane at the Turn - 0:08:39A.J. Brown Falls to Erickson - 0:09:04Trey McBride & Kyren Williams Build - 0:10:39Colston Loveland in Round 3 - 0:11:45Jeremiyah Love, Tetairoa McMillan & Josh Allen - 0:12:40Emeka Egbuka at 3.11 - 0:13:13Early Draft Heat Check - 0:14:13D-Bro’s RB-RB-TE Start - 0:15:05FantasyPros Giveaway - 0:15:58Garrett Wilson & Jaylen Waddle Picks - 0:16:47Breece Hall as Erickson’s RB2 - 0:17:09Terry McLaurin in Round 4 - 0:18:20TreVeyon Henderson Upside Pick - 0:18:54Mike Evans to D-Bro - 0:20:03Christian Watson Breakout Bet - 0:20:55Quinshon Judkins & DeAndre Swift Picks - 0:22:01Tucker Kraft Discussion - 0:22:37Parker Washington Upside Swing - 0:23:41David Montgomery & Jackson Dart Picks - 0:25:07Tony Pollard as an RB3/Flex - 0:26:07Trevor Lawrence Pick & QB Tier Drop - 0:27:02Trevor Lawrence Upside Case - 0:27:59BettingPros College Football Promo - 0:29:28Jadarian Price & Wan’Dale Robinson Picks - 0:30:40Rashad White PPR Value - 0:32:58Ricky Pearsall, Mekhi Lemon & Chris Godwin Picks - 0:34:38Jaden Higgins Upside Case - 0:36:07Jacoby Meyers & George Kittle Picks - 0:37:11Late Tight End Strategy - 0:37:59Kyle Pitts Value - 0:39:24Alvin Kamara’s New Value - 0:40:21Keaton Mitchell Late-Round Swing - 0:41:00Matthew Golden Upside Pick - 0:42:11Dak Prescott in Round 11 - 0:43:06Chig Okonkwo as a Late TE Pairing - 0:43:50Hard Rock Bet - 0:44:50Brock Purdy & Denzel Boston Picks - 0:47:01Tre Harris Late-Round Upside - 0:48:19Malik Willis Upside Pick - 0:49:58Emanuel Wilson Handcuff Pick - 0:50:31Stefon Diggs Late-Round Value - 0:51:00Greg Dulcich Tight End Depth - 0:52:00Draft Wizard Grades Revealed - 0:53:00Erickson Team Review - 0:54:00D-Bro Team Review - 0:57:00Bench Value vs. Starter Value Debate - 0:58:00Jake Ciely Team Review - 1:00:00Draft Wizard Pick Analysis Tool - 1:02:00Welsh’s Team Review - 1:04:00Draft Wizard Final CTA - 1:06:50Final Thoughts - 1:08:15Outro - 1:08:36 Helpful Links: Hard Rock Bet - Sign up for Hard Rock Bet and make a $5 bet and you'll get $150 in bonus bets if you win. Head over to Hard Rock Bet, sign up and make your first deposit today. Payable in bonus bet(s). Not a cash offer. Offered by the Seminole Tribe of Florida in FL. Offered by Seminole Hard Rock Digital, LLC, in all other states. Must be 21+ and physically present in AZ, CO, FL, IL, IN, NJ, OH, TN or VA to play. Terms and conditions apply. Concerned about gambling? In FL, call 1-888-ADMIT-IT. In IN, if you or someone you know has a gambling problem and wants help, call 1-800-9-WITH-IT. GAMBLING PROBLEM? CALL 1-800-GAMBLER (AZ, CO, IL, NJ, OH, TN, VA) Draft Wizard - Dominate your fantasy football draft with Draft Wizard. Run fast mock drafts, test different strategies, build custom cheat sheets, get pick-by-pick draft advice, and learn your leaguemates' tendencies before draft day. Just download the FantasyPros App or head to fantasypros.com/draftwizard Follow us on Twitch - The team here at FantasyPros is taking questions all week, every week on Twitch. Follow us on Twitch at twitch.tv/fantasypros and never miss a stream! Discord – Join our FantasyPros Discord Community! Chat with other fans and get access to exclusive AMAs that wind up on our podcast feed. Come get your questions answered and BE ON THE SHOW at fantasypros.com/chat Leave a Review – If you enjoy our show and find our insight to be valuable, we’d love to hear from you! Your reviews fuel our passion and help us tailor content specifically for YOU. Head to Apple Podcasts, Spotify, or wherever else you get your podcasts and leave an honest review. Let’s make this show the ultimate destination for fantasy football enthusiasts like us. Thank you for watching and for showing your support – https://fantasypros.com/review/ BettingPros Podcast – For advice on the best picks and props across both the NFL and college football each and every week, check out the BettingPros Podcast at bettingpros.com/podcast, our BettingPros YouTube channel at youtube.com/bettingpros, or wherever you listen to podcasts.See omnystudio.com/listener for privacy information.

Don't let the weather change the plan. Stormburst is built for unpredictable days, wet trails, and whatever the forecast throws your way.—grab 15% off here: https://vessi.com/wanshow • Free shipping • 30‑day returns • 1‑year warranty Sign up with the following links and enjoy the first 6 months of OpManager Nexus free. On-Site: https://lmg.gg/site24x7 Cloud: https://lmg.gg/manageengine Traditional carriers track your data. Cape protects it with built-in privacy features like IMSI rotation and last-mile encrypted texting. Use code WAN and get 33% off your first 6 months: https://cape.co/wan App Download Links: Apple: https://lmg.gg/CapeIOS Google: https://lmg.gg/CapeAndroid Accrescent App Store: https://accrescent.app/ Thanks Wispr Flow for sponsoring this video! Check them out at https://ref.wisprflow.ai/wan-show-yt-1 and use code WANSHOW for a free month of Wispr Flow Pro! Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

The middle rounds of your fantasy football draft can make or break your season.In this episode of No Punt Intended, we're breaking down the best and worst fantasy football values in Rounds 9-12 based on current 2026 ADP. These are the rounds where savvy fantasy managers find league-winning upside while others reach for players who carry far more risk than reward.Is Sam LaPorta one of the best values in Round 9? Has Kyle Pitts become an incredible bargain after years of disappointment? Are fantasy managers drafting Matthew Golden and Jordan Mason too early? We go round by round to identify the players we're targeting, the players we're fading, and how to maximize value throughout the middle rounds of your draft.Whether you're drafting in redraft, PPR, Half-PPR, or Superflex leagues, this episode will help you understand where the biggest ADP inefficiencies exist and how to build a roster with championship-winning upside.If you're preparing for your 2026 fantasy football draft, this is an episode you won't want to miss.If you feel like talking ball with us, come and join the Club Fantasy FFL/Women of Fantasy Football Discord!

Send us Fan MailTwo competitors deciding to stop fighting and build together always raises one question: what changes for the customers the next morning. We start with the BT and Verizon International joint venture and dig into what a shared global network footprint could mean across SD-WAN, MPLS, Ethernet and cloud connectivity, including the uncomfortable parts like orchestration changes, duplicated POPs and the possibility of forced migrations. If you run global sites, this is the kind of deal that can quietly reshape your WAN roadmap.Then we shift to the money and physics behind the cloud: Virginia's new per kilowatt-hour data center electricity tax. It sounds small until you apply it to hyperscale consumption, and the inclusion of self-generated power closes an obvious workaround. We talk through why Northern Virginia's data center corridor matters, how policy spreads state to state, and why the “they'll just pass the cost on” argument is less theory than a pricing strategy you eventually see in your cloud bill.From there, it's security and governance: the White House post-quantum cryptography executive order and the accelerating reality of quantum risk, including “harvest now, decrypt later.” We connect PQC deadlines to refresh cycles, vendor readiness, and the practical work of migrating both infrastructure and applications. We also hit identity head-on with Cisco's plan to bring identity lifecycle security into Splunk's agentic SOC, because non-human identities and AI agents are changing what least privilege even means.We close with the growing pattern of government intervention in frontier AI models and what it does to businesses building on specific model capabilities. If you found this useful, subscribe, share the show with a friend and leave a review so more people can find it.Check out the Monthly Cloud Networking Newshttps://docs.google.com/document/d/1fkBWCGwXDUX9OfZ9_MvSVup8tJJzJeqrauaE6VPT2b0/Visit our website and subscribe: https://www.cables2clouds.com/Follow us on BlueSky: https://bsky.app/profile/cables2clouds.comFollow us on YouTube: https://www.youtube.com/@cables2clouds/Follow us on TikTok: https://www.tiktok.com/@cables2cloudsMerch Store: https://store.cables2clouds.com/Join the Discord Study group: https://artofneteng.com/iaatj

EPISODE 443 - Not All Who Wan-der Are Lost: A (final) Journey Through The Wan-iverse PT. 5 Well we've done it. After 6ish years of ups and downs, we've finally conquered the mighty Wan-verse. Well, minus the Nun 2 which we honestly forgot even existed. Which is what will happen to this show in like a month. Or maybe we'll haunt you like a creepy mirror for generations.... THIS WEEKS MOVIE: The Conjuring: Last Rites THIS WEEKS BEER: Dragonmead Brewery - Final Absolution Belgian Tripel Follow us! Twitter: @thebuzzedkillPC Instagram: @thebuzzedkillpodcast Facebook.com/thebuzzedkillpodcast Letterboxd: Buzzed Kill Podcast

Visit https://www.squarespace.com/WAN and use offer code WAN for 10% off Thanks to MSI for sponsoring this video! Go to https://lmg.gg/InV81 and check out their MPG 322UR QD-OLED X24 Monitor! Pick up some TOSHIBA N300 Pro NAS Hard Drives today at https://lmg.gg/n300prowan (Product Info Site: https://lmg.gg/n300proofficial) Get an affordable sit-to-stand desk and elevate your work space! Check out the MotionGrey Ergo2 Pro at https://motiongrey.com/r?id=no5sc5 Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

Berî çend rojan DEM Partî bi navê 'Mîtînga Azadîyê' xwepêşandaneke cemawerî li darxist. Çareseriya demokratik a pirsgirêka Kurd û azadîya rêberê PKK peyamên sereke bûn. Ji aliyekî din ve partiyên Kurdistanî li Wanê kom bûn û nerazîbûna xwe ji siyaseta dewleta Tirkîyê, û ji sîyaseta entegrasyona demokratîk diyar kirin.

Enter AMD's latest giveaway at https://bit.ly/3RU8jkA to win a brand-new ROG Xbox Ally X! Check out Redtiger during their Mega Sale from June 23rd to June 26th and save up to 50% off site-wide! Grab the F7NA with Code: RTF7NA4K for a final price of $118.99. Amazon: https://lmg.gg/GwVQ6 Official Site: https://lmg.gg/mK6bG Check out the Shokz OpenFit Pro through our links below and take advantage of their Prime Day sale with up to 35% off. Use code “WAN” at checkout for an extra 5% off. https://beopen.shokz.com/WANSHOW Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

Sign up with the following links and enjoy the first 6 months of OpManager Nexus free. On-Site: https://lmg.gg/site24x7 Cloud: https://lmg.gg/manageengine Traditional carriers track your data. Cape protects it with built-in privacy features like IMSI rotation and last-mile encrypted texting. Use code WAN and get 33% off your first 6 months: https://cape.co/wan App Download Links: Apple: https://lmg.gg/CapeIOS Google: https://lmg.gg/CapeAndroid Accrescent App Store: https://accrescent.app/ Check out Redtiger during their Mega Sale from June 23rd to June 26th and save up to 50% off site-wide! Grab the F7NA with Code: RTF7NA4K for a final price of $118.99. Amazon: https://lmg.gg/GwVQ6 Official Site: https://lmg.gg/mK6bG Get an affordable sit-to-stand desk and elevate your work space! Check out the MotionGrey Ergo2 Pro at https://motiongrey.com/r?id=no5sc5 Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

In a chamber of the Castle Eschatonica, filled with technology from another world, Veile, Caoimhe, Elena, and Brontë investigate the truth behind their mysterious gems (and the seemingly lost universes they were once connected to). Meanwhile, on the other side of the metaphysical fortress, Uncle Nicky, Jonathan, Antistrophe and their newly acquired Agonies encounter a brusque soldier from a war whose rules they can barely understand. This week on Perpetua: Escape the Rumbling Castle! 03 Perpetua Guide [In Progress v.06] Celestial Echoes [CECH] Imago [CEIM] While I haven't played the Imago games myself, from what I've read, it's a game about the power and cost of the creative spirit. And also demons, maybe? In any case, the way that this Echo shows up in the Castle Eschatonica is the Agonies, a special set of NPCs you can acquire who can grant even non-spell casters the ability to cast a few key spells!  Melo (he/him) Traits: Relaxed, Friendly, Sleepy Status/Element: Slow/Air  Description: A living statue of a little knee high elephant man, smoking a pipe. Ability: Trumpet Toss  - 10 - Self - Instantaneous Melo wakes up and releases blasts of air from his trunk, sending you flying across the battlefield. You may immediately perform a free attack with a melee weapon you have equipped. This attack may target creatures that can only be targeted by ranged attacks. If you used a weapon belonging to the brawling or spear Category for this attack, it deals 5 extra damage. If you hit a flying target with this attack, you may force them to land immediately. Wan (he/him) Traits: Sickly, Curious, Hopeful Status/Element: Weak/Ice Description: A painting of a strange looking child in bed, under a blanket, covered in shadows. Ability: A Chill Air (attack) - 10 x T - Up to Three Targets - Instantaneous Wan coughs, and the temperature in the room drops low. The targets suffer【HR + 15】ice damage. Opportunity: Each target hit by this spell suffers Weak. Hezzi (they/them) Traits: Afraid, Tentative, Big Status/Element: (Shaken/Physical)  Description: A ghostly, one-eyed ogre who quietly plays their piano until someone approaches. Ability: Intimidating Presence (attack) - 5 x T - Up to Three Creatures - Instantaneous Hezzi shows themself in fully material form, strikes an aggressive pose, and roars, giving every target Shaken. Miss Mephitic (she/her)  Traits: Layered, Beautiful, Frightening Status/Element: (Poisoned/Poison) Description: A human-sized music box ballerina, from which vines, petals, and teeth emerge. Ability: Bloody Thorn (attack) - 20 - One Creature - Instantaneous Mephi's vines strike out at a target, doing 【HR + 15】poison damage and giving them the Poisoned status. Connie (she/they) Traits: Excitable, Mercurial, Confused Status/Element (Dazed/Bolt)  Description: A fuzzy television on a wheeled cart which continually changes channels, occasionally landing on young woman with frizzy, dirty blonde hair . Ability: Surprise Shock (attack) - 20 - One creature - Instantaneous Connie slams into your foe and electrifies them. The target suffers【HR + 25】bolt damage. Damage is doubled against a target who is Dazed Ira (he/him) Traits: Irritable, Small, Loyal Status/Element: Enraged / Fire Description: A ceramic pot with a little angry red devil cherub on it—and the impish being itself. Ability: Devil's Breath (attack) - 10 × T - Up to three creatures - Instantaneous Ira unleashes a blast of fiery breath at your foes. Each target hit by this spell suffers【HR + 15】fire damage. Opportunity: Each target hit by this spell suffers enraged. Hosted by Austin Walker (austinwalker.bsky.social) Featuring Ali Acampora (ali-online.bsky.social), Art Martinez-Tebbel (amtebbel.bsky.social), Jack de Quidt (notquitereal.bsky.social),  Janine Hawkins (@bleatingheart), Sylvi Bullet (@sylvibullet), Keith J Carberry (@keithjcarberry) and Andrew Lee Swan (swandre3000.bsky.social) Produced by Ali Acampora Music by Jack de Quidt (available on bandcamp) Cover Art by Ben McEntee (https://linktr.ee/benmce.art) With thanks to Amelia Renee, Arthur B., Aster Maragos, Bill Kaszubski, Cassie Jones, Clark, DB, Daniel Laloggia, Diana Crowley, Edwin Adelsberger, Emrys, Greg Cobb, Ian O'Dea, Ian Urbina, Irina A., Jack Shirai, Jake Strang, Katie Diekhaus, Ken George, Konisforce, Kristina Harris Esq, L Tantivy, Lawson Coleman, Mark Conner, Mike & Ruby, Muna A, Nat Knight, Olive Perry, Quinn Pollock, Robert Lasica, Shawn Drape, Shawn Hall, Summer Rose, TeganEden, Thomas Whitney, Voi, chocoube, deepFlaw, fen, & weakmint This episode was made with support from listeners like you! To support us, you can go to friendsatthetable.cash.

With The Punisher appearing in this week’s trailer for Spider-Man: Brand New Day, join us as we dive into the script that became Punisher War Zone, including a more brutal Jigsaw origin, a different fate for Microchip, and an alternate finale! Support Valerie Chow’s GoFundMe here: https://www.gofundme.com/f/fun-in-fk-cancer Ben Wan’s comics, Alter Ego and Musashi & Son: https://www.amazon.com/stores/Ben-Wan/author/B0CZBKYPF3 Website: http://www.benwanwriter.com YT Channel: @benwan4141   Scripted Podcasts Written by Ben Wan   Ben Wan’s voice acting as the Arkham Tapes Batman, Michael Keaton Batman, and Val Kilmer Batman: Ben Wan- Voice Acting   […] The post The Original Punisher War Zone Script by Lexi Alexander & Nick Santora appeared first on Multiverse Of Color.

Get a free 15-day trial of Odoo's all-in-one business solution and see how it can make your life easier! Check it out at https://www.odoo.com/wan Sign up with the following link for a free 30-day trial of OpManager Nexus: https://lmg.gg/manageengine Visit https://www.squarespace.com/WAN and use offer code WAN for 10% off Get the drives you need without waiting months for supply to catch up. https://www.serverpartdeals.com/wan Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

Horror history gets weird, reflective, and dangerously snackable this week as This Week in Horror History travels through June 8–14 with mischievous monsters, haunted hotels, supernatural comedy, demonic hauntings, creepy mirrors, and one killer dessert that may be eating America alive.This episode digs into a packed week of classic horror movies, cult horror sequels, supernatural blockbusters, 1980s horror, haunted-house horror, and strange consumer-satire body horror, including the release of Gremlins, the same-day arrival of Ghostbusters, the wider theatrical expansion of The Shining, the hit sequel The Conjuring 2, and the mirror-filled nightmare of Poltergeist III.Inside this episode:• Gremlins turns one cute little mogwai into small-town monster chaos and helps push Hollywood toward the PG-13 rating.• Ghostbusters brings haunted libraries, demonic possession, rooftop apocalypse, and paranormal exterminators into one of the biggest supernatural comedy franchises ever.• The Shining expands wider in U.S. theaters, pulling more audiences into Stanley Kubrick's cold, hypnotic nightmare inside the Overlook Hotel.• The Conjuring 2 sends Ed and Lorraine Warren to the Enfield haunting and introduces one of modern horror's most memorable demonic figures.• The Deep-Cut Spotlight goes to Poltergeist III, the strange 1988 sequel that traps Carol Anne inside a haunted Chicago high-rise full of mirrors, elevators, cold hallways, and unfinished supernatural business.Plus: a horror birthday roll featuring Johnny Depp, Natalie Portman, Jurgen Prochnow, and Adrienne Barbeau, a creepy look at how horror sneaks into family homes, hotels, corporate towers, grocery aisles, and blockbuster comedy, and a weekly recommendation for Larry Cohen's weird, gross, and sharply satirical cult classic The Stuff.From mogwai mayhem and ghost exterminators to haunted reflections, demonic forces, the Overlook Hotel, and a mysterious dessert that might be consuming its customers from the inside out, this week proves horror can invade almost anything: your home, your hotel room, your movie theater, your refrigerator, and your appetite.

The Tennessee Titans made a major commitment to building around quarterback Cam Ward this offseason, and fantasy football managers should be paying attention.With Brian Daboll taking over as offensive coordinator, Tennessee added new weapons in Wan'Dale Robinson and first-round rookie Carnell Tate, giving Ward the most talented supporting cast of his young career. But is this offense ready to take the next step? And can Tony Pollard continue to provide fantasy value as the centerpiece of the Titans rushing attack?In this episode of A Look Inside, the Club Fantasy FFL team breaks down the Titans' fantasy football outlook for 2026, highlighting the best draft values, breakout candidates, sleepers, fades, and players you need to know before your fantasy football draft.Topics include:• Cam Ward's fantasy football ceiling in Year 2• Carnell Tate's rookie outlook and immediate impact• Wan'Dale Robinson's fit in Brian Daboll's offense• Tony Pollard's fantasy football value in 2026• Titans sleepers, busts, and draft strategyWhether you play redraft, dynasty, keeper, or best ball fantasy football, this Tennessee Titans team preview will help you identify where the value lies before draft season heats up.If you feel like talking ball with us, come and join the Club Fantasy FFL/Women of Fantasy Football Discord!

Make summer plans easier with one pair that does it all! Vessi Weekend Classic and Stormburst are lightweight, and built for all-day comfort. Grab 15% off your first pair here: https://vessi.com/wanshow Free shipping • 30‑day returns • 1‑year warranty Visit https://www.squarespace.com/WAN and use offer code WAN for 10% off Thanks to MSI for sponsoring this episode! Head over to https://lmg.gg/asWi8 to check out their MPG 341CQR QD-OLED X36 monitor! Go to http://zapier.com/wan to join the millions of businesses transforming how they work with Zapier and AI Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

El programa destaca una trama del PSOE para proteger los intereses de Pedro Sánchez, según un informe de la Guardia Civil. Santos Cerdán y Leire dirigen esta "estructura criminal", utilizando la organización del partido. La UCO interpreta que las referencias al "WAN" aluden a Sánchez. Leire Díaz participa en reuniones en la Fiscalía General del Estado y ofrece pactos, presumiendo de contactos de alto nivel. La directora general de la Guardia Civil recibe a Leire para operar contra la UCO, y el DAO ordena a la UCO "ponerse de perfil" en asuntos que afectan políticamente al presidente, como el caso de su hermano. El juicio de David Sánchez avanza con la declaración del teniente coronel Balas, quien vincula a Miguel Ángel Gallardo como impulsor de la creación de su puesto en la Diputación de Badajoz, señalando que la decisión parte de un nivel superior. La visita del Papa a España, del 6 al 12 de junio, implica un despliegue de seguridad sin precedentes con 20.000 agentes y restricciones ...

As the northern edge of the Castle Eschatonica grows closer, the adventurers within begin to sense a change in the dungeon around them. While the Celestial Echoes that make up the hallways and rooms of the castle were once neatly separated, now it seems that the people and places from the various past Perpetuan worlds have begun to overlap and interact. In search of information, Brontë, Veile, Elena, and Caoimhe seek help from an otherworldly machine. Meanwhile, on the other side of the castle, a gang of otherworldly beings seeks help from the trio of Antistrophe, Jonathan, and Uncle Nicky.  This Week on Perpetua: Escape the Rumbling Castle! 02 Perpetua Guide [In Progress v.06] Some Feedback [Page 64 of 65] Doom_Tree_Anne  So, I HAVE to know… who is everyone's favorite Agony? Back in Imago 2: Ancestral Fault, I HATED Miss Mephitic, because I just thought compared to the other Agonies, she was a really boring design. Who cares about a bunch of vines!? But with the music box/ballerina re-design in this game… I think she might have jumped to the top of the list! I guess we have to wait and see if she keeps it going forward… TheUnforgivenIII My favorite is definitely Melo. That guy seems like he'd be cool to chill out with and listen to music with. Alukard83 Wow, color me surprised Unforgiven! I thought you'd be an Ira guy, through and through.  TheUnforgivenIII Why the hell would I like that stupid little idiot baby? Alukard83 Well, now I think it might be a bad idea to tell you why. As for my favorite, this is the first time I'm really meeting them, but I gotta say, I really love Connie. I love it when something references something else, you know? And she's sort of a walking (well, rolling) reference machine! TheDiamondRanger What's An Agony? The strategy guide says they're like equipable spells but it sounds like they're NPCS too!? CarlsSr You're such a Wan. Hosted by Austin Walker (austinwalker.bsky.social) Featuring Ali Acampora (ali-online.bsky.social), Art Martinez-Tebbel (amtebbel.bsky.social), Jack de Quidt (notquitereal.bsky.social),  Janine Hawkins (@bleatingheart), Sylvi Bullet (@sylvibullet), Keith J Carberry (@keithjcarberry) and Andrew Lee Swan (swandre3000.bsky.social) Produced by Ali Acampora Music by Jack de Quidt (available on bandcamp) Cover Art by Ben McEntee (https://linktr.ee/benmce.art) With thanks to Amelia Renee, Arthur B., Aster Maragos, Bill Kaszubski, Cassie Jones, Clark, DB, Daniel Laloggia, Diana Crowley, Edwin Adelsberger, Emrys, Greg Cobb, Ian O'Dea, Ian Urbina, Irina A., Jack Shirai, Jake Strang, Katie Diekhaus, Ken George, Konisforce, Kristina Harris Esq, L Tantivy, Lawson Coleman, Mark Conner, Mike & Ruby, Muna A, Nat Knight, Olive Perry, Quinn Pollock, Robert Lasica, Shawn Drape, Shawn Hall, Summer Rose, TeganEden, Thomas Whitney, Voi, chocoube, deepFlaw, fen, & weakmint This episode was made with support from listeners like you! To support us, you can go to friendsatthetable.cash.  

“When you need these systems, they have to work 100% of the time,” says Jake Jacoby, CEO of TELCLOUD. “Our solution doesn't just meet the old copper standard — it exceeds it.” In part 35 of the TELCLOUD POTS and Shots Podcast Series, Doug Green, Publisher of Technology Reseller News, speaks with Jacoby about the hardware architecture powering modern POTS replacement and why reliability remains the most important requirement for life-safety communications. The discussion focuses on TELCLOUD's purpose-built POTScast 8 and POTScast 2 devices, which support eight and two analog lines respectively. Designed specifically for POTS replacement, the units support applications including fire alarms, elevators, emergency phones, security systems, fax lines, SCADA systems, and other legacy communications still dependent on analog connectivity. Jacoby explains that traditional copper phone lines historically delivered both dial tone and power directly from the carrier's central office, making them highly reliable during outages. TELCLOUD's approach replaces that infrastructure with a more resilient, modern design featuring battery backup, multiple WAN paths, LTE and 5G connectivity, and remote monitoring capabilities. Each POTScast unit includes a built-in 24-hour battery backup with optional expansion capability, along with support for multiple WAN connections including fiber, satellite, and cellular. TELCLOUD also supports Power over Ethernet deployments, allowing cellular routers from providers including Digi and ATEL to be placed up to 250 feet away from telecom closets where signal strength is stronger. Jacoby noted that TELCLOUD originally relied on existing analog telephone adapters but ultimately engineered its own hardware platform after determining that available solutions did not meet the company's performance standards for mission-critical deployments. “These devices are designed to sit in that telco room for the next 20 years,” Jacoby said. The episode also explores how TELCLOUD combines hardware, platform services, monitoring, field services, and channel support into a fully managed POTS replacement offering delivered through reseller partners globally. The “Shots” segment of the podcast featured Casa 1560 Private Selection Extra Añejo, a tequila aged more than three years in oak barrels and described by Jacoby as having notes of dark chocolate, dried fruit, and oak. For more information, visit telcloud.com or call 844-900-2270.

Who are the wide receivers set to break fantasy football in 2026? On this episode of Club Fantasy FFL's No Punt Intended, we continue our annual Breakouts & Sleepers series by identifying the WRs poised to crush ADP, become league winners, and emerge as must-start fantasy football stars this season.Every year, fantasy managers miss on breakout WR production. Last season, players like Chris Olave and Wan'Dale Robinson were discounted by the fantasy community, while George Pickens showcased elite upside and Michael Wilson became a major value pick. This year, we're uncovering the next wave of fantasy football WR sleepers, breakout candidates, and undervalued wide receivers before the rest of your league catches on.In this episode:2026 fantasy football WR breakoutsLate-round WR sleepers with league-winning upsideUndervalued wide receivers to draftFantasy football ADP stealsWide receivers primed for career-best seasonsDraft strategy for redraft and best ball leaguesWhether you play redraft, keeper, best ball, or DFS, this episode is packed with fantasy football advice to help you dominate your 2026 drafts.Follow Club Fantasy FFL on YouTube, Spotify, and Apple Podcasts for rankings, mock drafts, dynasty strategy, rookie analysis, and fantasy football coverage all year long.If you feel like talking ball with us, come and join the Club Fantasy FFL/Women of Fantasy Football Discord!

Enter AMD's latest giveaway at Enter AMD's latest giveaway at https://bit.ly/3Oy0Iad to win an AMD Ryzen 7 9800X3D CPU+AMD Radeon RX 9070 XT GPU bundle! Traditional carriers track your data. Cape protects it with built-in privacy features like IMSI rotation and last-mile encrypted texting. Use code WAN and get 33% off your first 6 months: https://cape.co/wan App Download Links: Apple: https://lmg.gg/CapeIOS Google: https://lmg.gg/CapeAndroid Accrescent App Store: https://accrescent.app/ Get an affordable sit-to-stand desk and elevate your work space! Check out the MotionGrey Ergo2 Pro at https://motiongrey.com/r?id=no5sc5 Level up your streaming and recording game with XSplit Broadcaster! Get 30% off your first purchase or subscription with code WANSHOW30 at https://lmg.gg/xsplitwan Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

“These are purpose-built devices,” says Jake Jacoby, CEO of TELCLOUD. “They're UL listed, certified, tested, and designed specifically for this business.” In the latest episode of the TELCLOUD POTS and Shots Podcast Series, Doug Green, Publisher of Technology Reseller News, speaks with Jacoby about the hardware that makes modern POTS replacement possible. Jacoby showcases two TELCLOUD devices: the POTScast 8 LTE PC228 LTE, which supports eight analog lines, and the POTScast 2 LTE PC222 LTE, which supports two. Both are designed to support legacy and life-safety systems such as elevators, fire alarms, security systems, fax lines, SCADA applications, modems, and emergency phones as copper lines are phased out. The POTScast platform combines analog support with modern LTE and WAN connectivity, including broadband, Wi-Fi as WAN, satellite, and cellular. Each device includes 24-hour battery backup, helping ensure that critical communications continue even when building power fails. Jacoby also explains TELCLOUD's modular design. Because cellular signal is often weak inside telecom rooms, TELCLOUD supports Power over Ethernet, allowing routers from partners such as Ericsson, Peplink, Digi, InHand, ATEL, and Seego to be placed up to 250 feet away for better reception. The episode closes with the Shots segment, featuring Herencia Historico Grand Reserve Extra Añejo, a five-year-aged, small-batch tequila from Jalisco presented in a distinctive handcrafted bottle. For more information, visit telcloud.com or call 844-900-2270.

Thanks to Tello for sponsoring this video! Check out their plans at https://tello.com/? utm_source=partner&utm_medium=cpc&utm_campaign=WAN&src=partner&mdm=cpc&cmg=WAN Sign up with the following link for a free 30-day trial of OpManager Nexus: https://www.manageengine.com/it-operations-management/download.html Thanks to MSI for sponsoring this episode! Go to https://lmg.gg/ooUMY to get your MSI Aegis PC today! Visit https://www.squarespace.com/WAN and use offer code WAN for 10% off Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

One pair for everyday and everywhere. Vessi claim their Weekend Neo is waterproof, lightweight, and built for daily wear and travel. Grab 15% off your first pair here: https://vessi.com/wanshow • Free shipping • 30‑day returns • 1‑year warranty Get a free 15-day trial of Odoo's all-in-one business solution and see how it can make your life easier! Check it out at https://www.odoo.com/wan Visit https://www.squarespace.com/WAN and use offer code WAN for 10% off Start your free trial with NinjaOne today: https://see.ninjaone.com/LinusTechTips Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

The WAN Show will be slowly moving to the https://www.youtube.com/@WANShow channel! Check out this episode here: https://youtu.be/MwIb0ATb2cc One pair for everyday and everywhere. Vessi claim their Weekend Neo is waterproof, lightweight, and built for daily wear and travel. Grab 15% off your first pair here: https://vessi.com/wanshow Free shipping • 30‑day returns • 1‑year warranty Visit https://www.squarespace.com/WAN and use offer code WAN for 10% off Purchase the Shokz OpenFit Pro using the link: https://beopen.shokz.com/WAN-openfitpro Thanks to Tello for sponsoring this video! Check out their plans at https://tello.com/?utm_source=partner&utm_medium=cpc&utm_campaign=WAN&src=partner&mdm=cpc&cmg=WAN Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

Take a Network Break! Our Red Alert covers a trio of vulnerabilities in Cisco ISE. On the news front, Cloudflare announces a private network offering for AI agents and a partnership with CNAPP specialist Wiz for AI visibility. AWS rolls out Interconnect to streamline provisioning of WAN and last-mile connectivity, and Linux 7.0 includes network... Read more »

Take a Network Break! Our Red Alert covers a trio of vulnerabilities in Cisco ISE. On the news front, Cloudflare announces a private network offering for AI agents and a partnership with CNAPP specialist Wiz for AI visibility. AWS rolls out Interconnect to streamline provisioning of WAN and last-mile connectivity, and Linux 7.0 includes network... Read more »

Last time we spoke about the beginning of the first battle of Changsha. From Chongqing, Chiang debated defensive strategies for Hunan, ultimately adopting Plan B after Xue Yue's pleas, focusing on successive resistance north of Changsha to thwart Japanese advances. Japanese forces, under Okamura Yasuji, launched assaults in Jiangxi and Hunan. In Jiangxi, the 106th and 101st Divisions attacked Huibu and Gao'an, where Chinese troops under Luo Zhuoying and Song Kentang fiercely resisted. Gao'an fell briefly but was recaptured by the 32nd Army and the elite 74th Army, with heavy casualties on both sides, as recounted by soldier Liu Qihuai. In Hunan, Japanese units crossed the Xin Qiang River and landed at Yingtian, facing brutal opposition. At Bijia Mountain, Qin Yizhi's 195th Division held for four days; Battalion Commander Shi Enhua's reinforced unit perished entirely, their fragmented remains mourned by locals. Along the Miluo River, Chen Pei's 37th Army fortified positions, repelling waves of Japanese attacks, including suicide squads disguised as civilians. Recruit Yang Peyao's unit endured bombardments, inflicting significant enemy losses before withdrawing at dusk.   #197 The First Battle of Changsha Welcome to the Fall and Rise of China Podcast, I am your dutiful host Craig Watson. But, before we start I want to also remind you this podcast is only made possible through the efforts of Kings and Generals over at Youtube. Perhaps you want to learn more about the history of Asia? Kings and Generals have an assortment of episodes on history of asia and much more  so go give them a look over on Youtube. So please subscribe to Kings and Generals over at Youtube and to continue helping us produce this content please check out www.patreon.com/kingsandgenerals. If you are still hungry for some more history related content, over on my channel, the Pacific War Channel where I cover the history of China and Japan from the 19th century until the end of the Pacific War. Major Luo Wenlang, battalion commander of the 3rd Battalion, 55th Regiment, 19th Division of the 28th Army, harbored a peculiar quirk: he couldn't sleep soundly without unwrapping his leg bindings, a small ritual that anchored him in the chaos of war. Since the war's eruption, such luxuries were rare, and unwrapping his bindings every night became an impossibility, leaving him to endure restless slumbers. Tonight, however, sleep eluded him entirely; he tossed and turned on his makeshift bed, his mind a whirlwind of unrest. Two days after the northern Hunan battle ignited like a powder keg, the 55th Regiment received urgent orders from Division Commander Tang Boyin to race to Wukou in Pingjiang County. Their path wound through Luo Wenlang's hometown of Fulinpu, a twist of fate that stirred conflicting emotions. Entering the village under the cover of night, the entire battalion encamped in the commander's modest family village, with battalion headquarters naturally established in his ancestral home.   Luo yearned to step across that familiar threshold but dreaded it, for his parents remained oblivious to a devastating truth. They slaughtered chickens and prepared meat, hosting the battalion staff with drinks and hospitality, after all, this was their son's unit gracing their home. Luo orchestrated door planks and straw for bedding, posted sentries, and deftly evaded his parents until they retired. Before dawn broke, he mustered the troops, ensured they were fed, and led them onward, slipping away like a shadow. By noon on the 22nd, they reached Wukou, only to receive fresh directives: rush to Yingtian to bolster the 95th Division against the enemy's audacious landings. The 3rd Battalion spearheaded the division's reinforcements, marching relentlessly through day and night, arriving at Dongtang, over 30 kilometers southeast of Yingtian—on the 23rd, hearts sinking upon learning Yingtian had already fallen into enemy clutches.   Luo Wenlang sought out the retreating 95th Division Commander Luo Qi to beg for a mission, his resolve unyielding. Luo Qi, anticipating his arrival, relayed Commander Guan Linzheng's ironclad instructions: The 19th Division's reinforcements would assume Dongtang's defenses. With the main force still en route, Luo Qi tasked Luo's battalion with relieving a segment held by a replacement regiment. He handed over a map, sketching a line with a pencil, a simple stroke that thrust Luo Wenlang and his men onto the front lines of fate. An operations staff was dispatched to guide them to the position and oversee the handover.   As the troops advanced, they encountered scattered soldiers fleeing like startled rabbits; seizing a platoon leader revealed they were indeed from the replacement regiment. Mere minutes from division HQ, the enemy was already closing in, a predator's breath hot on their necks. Luo Wenlang and Deputy Battalion Commander Wu Yacui split the battalion, launching a counterattack on Dongtang from dual routes. Fortune favored them; the Japanese held only an exhausted company, crumbling under a single, ferocious charge. They swiftly deployed two companies to the positions, reserving one as a bulwark. By dusk, the full 55th Regiment arrived, accompanied by the rest of the 19th Division's reinforcements, allowing the battered 95th Division, ravaged at Yingtian, to withdraw for desperate reorganization. The regimental commander positioned Luo's 3rd Battalion on the regiment's vulnerable left wing. In the blink of an eye, it was the 27th, aligning with the 15th of the eighth lunar month. Amid the relentless great battle, few noted the calendar, and the skies hung heavy with clouds. Luo Wenlang twisted on his straw bed, his thoughts a snarled knot of anxiety and memory.   At 11 p.m., gunfire shattered the night; a barrage of machine gun bullets riddled the battalion HQ house, raining thatch and dust upon Luo like fallout from a storm. Catastrophe had struck! Luo surged toward the positions with the bugler—his battalion signal chief—and the reserve force, ascending the hilltop in a frenzy. Halfway up, he spotted 8th Company's Lieutenant Platoon Leader Rong Fayu leading over 20 soldiers in retreat. Bellowing "Why unauthorized retreat?" while brandishing his pistol, he compelled Rong to rally and turn back. The Japanese had launched a nocturnal assault; 8th Company Commander Yi Zuitao lay slain by a fatal shot, over a dozen comrades felled in brutal close combat, the survivors scattered like leaves in the wind; the high ground now belonged to the enemy.   Upon learning of Dongtang's loss, the regimental commander personally led the regimental reserve, his face etched with urgency. Under flickering lantern light, poring over the map with Luo, Division Commander Tang Boyin telephoned, his voice a whipcrack of command: Recapture it before dawn, or both would face the merciless hand of military justice. After seizing the high ground, the enemy hesitated to press further; Luo surmised the darkness concealed paths, and their numbers were not overwhelming. Forgoing the regimental reserve, he led 7th Company's 4 squads and remnants of the routed 8th Company in a stealthy ascent. Near the position, a ravine concealed over 20 8th Company soldiers, rallied by Sergeant Squad Leader Tan Tianrong, who had lurked in wait for reinforcements, dreading exposure at dawn under the enemy's gaze.   Spotting the battalion commander personally spearheading the counterattack, Tan Tianrong's face lit with fierce joy; his men, armed with grenades, surged as the vanguard. Intimate with the terrain even in blindness, they hurled explosives into bunkers, trenches, and works. The commander orchestrated the charge; the Japanese force of 40-50 men crumbled, over half slain or maimed, the remnants fleeing northward to their village stronghold. It was past 4 a.m.; the moon pierced the clouds, bathing the earth in a silvery glow. With positions reclaimed, the night revealed its secret: tonight was Mid-Autumn. Moonlight unraveled the tangled threads of his past; Luo draped his clothes over his shoulders, sat beneath the luminous orb, and wept in solitary anguish.   Before the war, devastating news had arrived: his brother Luo Yinong had been killed in Jiangxi. Luo had three brothers; the eldest shouldered half the family's burdens, their bond unbreakable. The brother had enlisted first in the 50th Army, climbing to battalion commander through sheer valor. He and his younger brother had followed suit, inspired by that call to arms.   Wartime conscription demanded only one per family, but battling the devils was a duty for the nation and its people. His brother had risen to deputy regimental commander before his end. The 50th Army notified him first. Engulfed in battle, there had been no time to console his grieving parents or tend to the funeral; it weighed on his heart like an unyielding stone. His sister-in-law, diligent and unassuming, cared for a young boy and carried another child; the long, arduous days ahead loomed like an endless shadow. The night dew brought a biting chill, the moon an icy sentinel; Luo shivered uncontrollably, his tears mingling with the frost.   The sky hung heavy with overcast gloom, yet the moon lurked beyond the clouds, casting a faint, ethereal light that warded off utter darkness. Along the road, a unit's elongated black shadow snaked southward in hurried silence, a serpent of weary resolve pressing through the night. Qin Yizhi reined in his horse, pausing to gaze back: the queue stretched onward, silent and impeccably orderly, belying the exhaustion of a force scarred by days of ferocious combat, their spirits unbroken amid the shadows. After the Japanese seized the 195th Division's defiant outpost at Bijia Mountain, they surged across the Xin Qiang River in a merciless onslaught. The river, shallow enough to wade knee-deep, offered no true impediment; the real barrier was forged from the defenders' scorching blood, a crimson testament to their unyielding stand. The 195th Division clashed in a maelstrom of cruelty; positions were heaped with corpses time and again, the Xin Qiang's waters churning blood-red in relentless cycles of carnage. From the night of the 23rd to the dawn of the 25th, respite was a forgotten dream; Okamura Yasuji, in a gesture of grim respect, inscribed Qin's name in elegant calligraphy and hung it within his command tent, a haunting trophy of the foe's tenacity.   Following their triumphant landing at Yingtian, the Japanese entangled the Ninth War Zone's left-wing defenders in a protracted snare, their advances grinding slowly like a predator toying with prey, menacing the flanks of the frontal troops with insidious intent. On the evening of the 27th, Xue Yue issued the fateful order for the 15th Army Group to withdraw to the precarious ground between the Miluo River and Shangshan City, ushering this blood-soaked force into an all-night march toward the next defensive crucible. Late into the night, a brief halt was called. Soldiers slumped to the ground, adjusting leg wraps and gear with mechanical precision; logistics teams darted through the ranks, distributing rations like lifelines; cooks, having forged ahead, arrived with steaming pots of rice soup, infusing the air with a rare warmth. Though no clamor broke the hush, a quiet camaraderie enveloped the queue, a fleeting balm against the war's chill.   The division staff claimed a flat expanse beside a farmhouse yard for their respite. Qin settled onto a stone roller used for grinding grain, nibbling at his meager ration and sipping the hot soup that steamed in the cool air. Suddenly, moonlight pierced the clouds, cascading down in silvery streams; the familiar contours of the farmhouse stirred a flood of warmth in his heart, evoking memories of home.   Chongqing, Huangshan Villa. Every window was shrouded in double layers of thick curtains, sealing out any sliver of betraying light, as if the very walls conspired to guard secrets from the encroaching night. Tonight's ethereal protagonist rose languidly from the eastern valley, its orange-red moonlight casting an aura of drowsy reluctance, as though it had not fully shaken off the slumber of the day. The feeble glow dappled the building's roof, balcony, and the surrounding hillsides, intersections, and thickets, where armed shadows lurked, capturing every rustle in the oppressive silence. Only upon close inspection could one discern the faint specks of moonlight glinting off steel helmets. Yet, beyond those fortified walls, another realm pulsed with life, a vibrant contrast to the shadowed vigilance outside. The front hall, living room, and dining room blazed with brilliant light. Vibrant flowers, dominated by chrysanthemums in full, defiant bloom, infused the air with color and fragrance; a phonograph murmured a cheerful Guangdong melody, weaving an atmosphere thick with festive joy, a deliberate illusion amid the storm of war.   Chiang Kai-shek, clad in a flowing black silk gown, strode ahead with poised grace, escorting his guests into the dining room alongside the elegantly attired Soong May-ling, their conversation laced with laughter and warmth. At the table, Soong May-ling's smile was a beacon of diplomacy, as she artfully arranged the seating to suit hierarchies and alliances, while servers in crisp white uniforms moved with nimble precision. This was Chiang Kai-shek's intimate Mid-Autumn family banquet; beyond a handful of pivotal military and political figures, the gathering brimmed with relatives. Guests and kin alike noted Chiang's buoyant spirits tonight; his smiles were wide and genuine, his discourse light and expansive, delving into casual topics with uncharacteristic ease.   In September 1939, China's War of Resistance Against Japan had entered its grueling third year. After the initial cataclysm of turmoil and disarray, the government and military had clawed their way to stability, adapting to this unprecedented historical crucible, with operations finally aligning into a semblance of order. According to figures proclaimed by Minister of Military Affairs He Yingqin to Chinese and foreign reporters on the 13th of this month, Japanese invaders had seized 521 counties across 12 provinces, a vast swath of conquest. Yet, the Japanese imperialists had exacted this toll at a staggering cost. Just prior, on August 30, the Hirannuma Cabinet, installed a mere eight months earlier, had collapsed in mass resignation. Hirannuma Kiichiro's predecessor, Konoe Fumimaro, had similarly bowed out amid governmental failures, chiefly the unmet ambitions in the Sino-Japanese War that he had boldly promised to parliament, exacerbating domestic political and economic woes. Days ago, when Wang Pengsheng briefed Chiang on Japan's turbulent politics, he quipped: "Konoe said three months to destroy China; three months didn't work, nor three years, who knows about 30 or 300. Hirannuma had no solutions, down in eight months. Does Abe have good ideas? How long can he be prime minister?" Indeed, Abe Nobuyuki, Hirannuma's successor, would endure a mere four and a half months before resigning in ignominy. Tonight's feast showcased Chiang's favored cuisines: delicate Jiangsu-Zhejiang dishes mingled with robust Sichuan flavors. Chiang abstained from alcohol, raising his cup in mere symbolic toasts to his guests. During the meal, as if by unspoken accord, no one broached the raging domestic battles or the volatile international landscape; conversations meandered through trivialities, skirting anything heavy or discordant, a fragile bubble of normalcy.   On September 3, Britain and France had declared war on Germany, shattering the global order in a seismic shift. Foreign newspapers already bandied the term "Second World War," a phrase that evoked freshness, exhilaration, and sheer terror in equal measure. China's diplomacy surged with newfound vigor. In April, Ambassador to the US Wang Zhengting had negotiated a $20 million loan with American banks on China's behalf. In May, Stalin responded to Chiang's overtures, agreeing to exchange arms for Chinese tea, wool, raw hides, and more. A month later, the first consignment of light and heavy weapons—including artillery and heavy machine guns—arrived via clandestine routes through Xinjiang and Mongolia, bolstering the central army's frontlines. In August, Hu Shih, Wellington Koo, and Chien Tai represented the Nationalist Government at the 19th League of Nations Assembly, laying bare the Japanese imperialists' atrocities in China before the world and rallying global forces for peace to support China's defiant stand. Soon after, British and American civic groups ignited "China Week" campaigns, pressing their governments to aid the beleaguered nation. Waves of foreign volunteers streamed in from distant shores: doctors, journalists, ordnance engineers, even retired soldiers clamoring to join the fray on the frontlines.   "If we could pull America into this war..." Through Soong May-ling's subtle, persuasive influence, Chiang allowed himself to daydream of that prosperous, dynamic young powerhouse across the vast ocean. Thus, on this Mid-Autumn night, his talk turned to America, to his correspondence with President Roosevelt regarding the "tung oil loan." That saga had unfolded the previous October; T.V. Soong had jetted to America, securing a loan with China's tung oil, a commodity scarce in the US, as collateral. China had boldly requested $400 million; America countered with $25 million, a classic tale of "ask high, settle low." Yet, the funds were secured. One success paved the way for many. Soong May-ling had once confided to Chiang: "In mobilizing US aid for China's resistance, I'll make a difference." When Chiang responded with a smile, "Thank you, Madam," he could scarcely foresee how his beautiful wife's extraordinary prowess in fulfilling this solemn vow would astonish him, etching eternal glory for Chinese women worldwide and elevating Soong May-ling to the zenith of her life's achievements.   The most direct echo of the First Battle of Changsha's thunderous saga resides in the Ninth War Zone's meticulous report on the northern Hunan and southern Hubei operations, submitted to the Chongqing Military Committee and Chiang Kai-shek himself, a faded relic now entombed amid the vast ocean of Nationalist Government military and political archives in Nanjing's Second Historical Archives of China. This document, a painstaking compilation of combat dispatches from divisions, armies, and army groups, stands as a testament to valor and sacrifice. Tragically, time's relentless march and human folly have ravaged this priceless artifact, leaving only shards and whispers to conjure the heart-wrenching inferno of that bloody clash.   "October 24, Year 28. Urgent. To Chongqing. Chairman Chiang. Secret. Submitted by Commander Xue on orders." The rice paper has yellowed to a deep, somber hue, brittle and parched; a careless touch could reduce it to dust. Some pages lie fractured, their remnants affixed to white paper, forever unable to reclaim their original wholeness. Leafing through page by page unleashes a pungent miasma, a scorched, acrid, decayed blend that assaults the senses. Traces of fire and water mar the original rice paper sheets, with countless fragments glued haphazardly to white backings, their sequences lost to eternity.   "...The Xin Qiang River spanning from Lujiao to Leishi Mountain, defending a front of over 110 li..." "Enemy 13th and 33rd Divisions, parts of the Hata Detachment, naval units, and artillery, cavalry, engineers totaling..." "...Began attacking us first with artillery... fortifications completely destroyed, then infantry charged; relying on our officers and men all resolved to coexist with the homeland..." "...And launched balloons to direct artillery... our army braved the cannons... repelled them, corpses filling the river, turning the water red..." "Division casualties also reached over a thousand... failed to inflict greater strikes and annihilate... deep inner guilt, besides vigorously training troops awaiting orders to kill the enemy..." "...Attack casualties heavy, then concentrated large forces... artillery fire so dense like continuous firecrackers for hours... released poison gas, Wang Street garrison all heroically sacrificed, then breached... Zhao Gongwu kowtows, October 15"   Zhao Gongwu commanded the 2nd Division under Zhang Yaoming's 52nd Army. This unit first held the line along the Xin Qiang River, then fell back to northeast of Fengjiang Bridge to staunch the enemy tide once more; after October 6, it hammered southward-marching Japanese from the west in the Yanglin Street and Dajing Street regions. Through these crucibles, the division bled over half its strength. A fragment of an envelope clings to a sheet of white paper, its words faintly visible: "Changsha 126-3 Zhang Yaoming," "Hunan Jinjing Air Mail," "Combat Process by..." and the like. The stamp remains remarkably intact—a philatelic gem now. Measuring 1.5 cm square, it features Sun Yat-sen's portrait at its center, inscribed "Republic of China Post" below, with "5" in the upper right, "fen" to the left, and "5" in each lower corner. I sat at the long table in the spacious, brightly lit reading room, staring vacantly, my thoughts grinding to a halt. These remnants are all that endure for posterity, of that monumental battle, of the scorching blood and vanished lives of countless unnamed Chinese soldiers. With hands that once gripped a rifle, I gently caressed those pages from a bygone era; they were cold, devoid of any lingering breath.   As the full moon of the 15th of the eighth month dissolved into the golden-red blaze of sunrise, Qin Yizhi's 195th Division had already plunged into the rugged mountains and dense forests encircling Fulinpu. Per directives from 15th Army Group Commander Guan Linzheng, the 195th was to forge a new defensive bastion centered on Fulinpu, 40 to 70 kilometers from Changsha. Their mandate: stall the Japanese southward juggernaut, granting precious time for allied forces to muster and fortify around the city. Despite the grueling all-night march, morale soared undimmed. The advance chief of staff doled out positions to each regiment, and the troops dove into fortification labors with fervent zeal. The 195th Division's unyielding stand along the Xin Qiang River had already etched preliminary glory upon this unit in its baptism of fire. "Fame in one battle" echoed as a battle cry throughout the division, where collective honor intertwined with personal valor. Honor and triumph formed the bedrock for soldiers and armies alike. Yet, another fire fueled their resolve.   On September 23, amid the Japanese forcing the Xin Qiang River, Guan Linzheng's voice crackled over the phone to Qin Yizhi: "Facing you is the 6th Division." The 6th Division, a name that ignited fury in Chinese troops and civilians, forever linked to the demonic specter of Tani Hisao. Moments later, the whisper spread like wildfire through every trench: "The Japanese army that perpetrated the Nanjing Massacre is right in front." Agitation rippled through the ranks; some donned fresh uniforms and shoes from their packs, casting aside the worn; others flouted discipline to bid farewells to hometown comrades: "Today we fight to the death here; see you in the next life." "Tell my mother I died fighting the Nanjing Massacre enemies."   Some company commanders commanded their mess sergeants to expend all funds on hearty feasts. All Japanese were foes, but the 6th Division embodied a blood debt, an unforgivable vendetta; the Chinese nation does not lightly forget its tormentors. In the Xin Qiang River maelstrom, the 195th Division battled with heroic ferocity. Some soldiers, in their final breaths, murmured: "Die then; it's worth it." Others lamented slaying too few devils, gritting teeth, eyes refusing to close in eternal regret. Now under Inaba Shiro's command, the 6th Division splintered southward after breaching the Xin Qiang; roughly a thousand hounded the 195th to Fulinpu. On the morning of September 29, the Japanese blundered into the 195th's meticulously laid ambush. Qin Yizhi, pulse racing with excitement and tension, fumbled the binoculars from his guard's hand. His command sliced the air: "Begin." War history chronicles: "The 6th Division advanced south from the Miluo River along the Xinshi-Liqiao road and Xinshi-Fulinpu routes. The over a thousand reaching Fulinpu were ambushed by the Nationalist 195th Division, suffering heavy losses." As Japanese artillery and aircraft unleashed hell upon the 195th's positions, Qin orchestrated a swift southward withdrawal to the environs of Shangshan City. Again, without pause, they erected fortifications and set deadly traps.   On the morning of September 30, the pursuers from Fulinpu closed in on Shangshan, their numbers swollen to over 1,500. Qin Yizhi clenched his jaw, his demeanor icy calm, allowing the Japanese to creep into the kill zone before barking: "Hit them hard!" Combat raged from dawn to dusk, obliterating over 700 foes. Qin ascended a hill, surveying through binoculars, then erupted: "Bad! The enemy is retreating." Upon receiving Qin's telegram, Guan Linzheng scrutinized the map, momentarily stunned, then replied: "Enemy shows no retreat signs yet; proceed per original plan. Your unit to block at Shangshan City line until October 2." Xianning, Okamura Yasuji's 11th Army HQ.   Combat maps bristled with markings, staff officers darting amid ringing phones and clattering telegrams. The colossal red arrow in northern Hunan had fractured into tributaries, surging over 100 km southward from the outset; one tendril pierced to Yong'an City, a mere 30 km from Changsha. Vast swaths of northern Hunan lay conquered, yet Okamura sensed the tide turning, it was time to retreat. The Chinese employed their time-honored gradual resistance, battling while retreating with cunning grace. Some units fell back directly, others amassed on flanks—what portent did that hold? In Okamura's shrewd mind loomed an equally shrewd Xue Yue; he envisioned his adversary methodically weaving a snare.   Post-Yingtian landing, the 15th Army Group's timely evasion had unraveled his "Xiang-Gan Operation Plan" like fragile thread. If encircling and annihilating the Chinese main force proved unattainable, what purpose in pressing onward? Telegrams from 3rd Division's Fujita Susumu, 6th's Inaba Shiro, and 13th's Tanaka Seiichi piled on his desk, pleading to assault Changsha—for headlines and Imperial accolades, perhaps, but blind to their exposed supply lines vulnerable to enemy thrusts? Ground logistics teetered on collapse; the air force resorted to airdrops for isolated regiments. Venturing further south would stretch lines to breaking; a severed artery spelled doom for the vanguard. When would these commanders mature into true stewards of the Imperial Army? Okamura fretted and pitied them in equal measure.   At 4 p.m. on September 30, Okamura decreed a halt to advances at Shangshan and Yong'an. He commenced orchestrating the retreat. Changsha, Yuelu Mountain, Ninth War Zone Command Forward HQ. October 1. Xue Yue stood before the map, Guan's latest telegram clutched in hand. Qin's second missive insisted on Japanese withdrawal, corroborated by 15th Army Group scouts from Yingtian: This morning (October 1), Japanese transports unloaded artillery stowed the previous night, hauling it back to Yueyang; intercepted wires revealed a regiment aborting its southward push, standing idle. Guan assessed the mosaic and commanded counteroffensives: intercept if feasible, pursue relentlessly, deny the Japanese escape; he relayed retreat indicators to Xue. Xue paced the chamber, head bowed in contemplation. Chief of Staff Wu Yizhi, Staff Director Zhao Zili, and their cadre tracked his every step with expectant eyes, awaiting the verdict. Xue's thoughts whirled through military stratagems and beyond.   Pre-war, Xue had segmented the war zone's forces into tripartite blocs: Northern Hunan under Guan Linzheng's 15th, Yang Sen's 27th, and Shang Zhen's 20th Army Groups as "A Cluster"; Northern Jiangxi Nanchang with Yunnan Army Lu Han's 1st Army Group and the 74th Army as "B Cluster"; the Wuning, Xiushui, Hunan-Hubei-Jiangxi border guarded by Sichuan Army Wang Lingji's 30th Army Corps, Fan Songpu's Border Advance Army, and 8th Army; augmented by 3 armies' 7 divisions in general reserve. Before the storm broke, Xue pored over maps, tracing every mountain, river, road, and bridge, envisioning burial grounds for the invaders.   Now, beneath Changsha, 200,000 troops formed a tightening net. The "decisive battle in Changsha suburbs" blueprint had been wired to Chongqing. Chiang and the nation yearned for a resounding triumph as the resistance pivoted into a new epoch?! A masterful drama, honed over half a month's toil, neared its crescendo; yet that cunning fox appeared to sniff the trap's metallic tang, freezing in place. "Commander, phone from Minister Chen." "Brother Boling, good news." Chen Cheng's voice brimmed with levity, "Your formal appointment published. What? Ninth War Zone Commander! First to congratulate; document tomorrow." Shedding the "acting" prefix was inevitable; Chiang had intimated as much long ago. But for a man and general, true worth lay not in titles, but in forging indelible feats. Splendor was judged not by underlings, colleagues, or superiors, but by peers in the craft of war.   Unmoved by the promotion, Xue exhaled a profound sigh. Though the 15th's intelligence couldn't confirm a wholesale retreat, preparations for dual contingencies were imperative. Victories came hard; a splendid battle, harder still. He summoned Wu Yizhi and Zhao Zili to devise countermeasures for the enemy's potential flight. October 2, Sichuan Army Yang Sen's 27th Army Group, Yang Gancai's 134th Division special service company, under Company Commander Wan Mingyu, slogged through the profound mountains and forests on the northern Mufu Mountains' flanks. The 134th's covert mandate: infiltrate enemy rear via treacherous terrain, sabotage supply arteries in the Chongyang-Xianning sector, and deliver a dagger to the Japanese spine when opportunity struck, bolstering frontal defenses.   Past 3 p.m., a crystalline mountain stream materialized. Wan decreed a respite. Over 100 soldiers, drained from a half-day's ascent, collapsed like puppets with severed strings. Most propped their torsos with rifles in one hand, fanning hats to ward off the relentless forest mosquitoes with the other. Regaining breath, they devoured rations washed down with stream water. Some unfurled towels and ventured downstream, letting the cool flow rinse away layers of sweat. Then, a muted engine drone encroached from the heavens. Wan peered through the foliage: a low-flying plane vectored southward, its wings emblazoned with the Rising Sun.   A transport; Wan recognized the temporary Japanese airfield near Xianning. With lines overextended, airdrops sustained isolated units. Wan was prying open a can with his bayonet, the tip etching a cross on the lid before levering along the edge; paired with a rice ball, it promised a savory repast. His orderly proffered a cup of fresh stream water; 2nd Platoon Leader Hu Yaozong perched nearby on a rock, smirking, poised to pilfer from the opened tin. Wan warded off this Sichuan Pixian compatriot. The plane droned overhead then.   Both glanced skyward; the platoon quipped: "Open quick, damn, I'll repay two cans later." Commander: "Want cans? Sky has; shoot plane down, enough for two lifetimes, bloat your mother-in-law first." The can hailed from a prior supply raid. Platoon: "You want me to shoot the plane?" Commander: "Bastard! You shooting or not?" The platoon snatched the light machine gun from a tree fork, jamming the butt against his belly, one hand on the grip, aiming crudely: "Come down, you turtle son!" The other hand squeezed the trigger. Wan assumed jest, resuming his task. "Da-da-da..." Wan jolted; the half-opened can tumbled to his feet, spilling Japanese fish onto Chinese soil. Recoil floored the platoon; he hurled the gun like a branding iron, face ashen. Inspecting the trigger, he snarled: "Whose damn fault, why no safety?!" The gunner dashed over; tall and even-tempered: "Safety was on; how'd it fire without pulling?" Wan's initial panic: "Damn! Position exposed."   The company spearheaded the division's reinforced regiment to raze a recent Japanese depot, guarded by a mere company—but exposure doomed the regiment deep in hostile territory. The assault had been plotted for days; pre-departure, Yang Gancai had toasted them. Wan had sworn a blood oath: No return to Sichuan without success. Hu had jested then: "No Sichuan return means wanting Hunan girl as concubine." Banter was fine in peace, but in war's grip, this was no trifling errand. Wan unleashed a torrent of curses, rising to survey the environs. The main force lagged 15 km behind; advance or abort post-blunder? Enemy rear was a labyrinth; this isolated band teetered on a razor's edge. As if to compel a choice, the radio operator approached; Wan itched to lash out. In his fury and indecision, a miracle unfolded.   The transport's engines hacked like a consumptive invalid, then a witness spied the plane banking left, plummeting, its nose inexorably toward a colossal rock 3-4 km distant. It rebounded twice on the stone, nose and left wing crumpling; the fuselage, fragile as parchment, tumbled gently, skewing onto the slope amid splintered trees. Wan gaped, then bellowed: "Assemble!" The men snapped from reverie, charging downhill in a frenzied cascade. One hour later, 134th Deputy Commander and Reinforced Regiment Commander Liu decoded Wan's vanguard transmission via radio. Another hour passed before Liu received Yang Gancai's directive: Abort Mountain Leopard operation; return with documents expeditiously. One day hence, October 3, Okamura Yasuji's original retreat order from October 2 dawn, addressed to northern Hunan's 6th, 33rd Divisions, Nara and Uemura Detachments, plus its Chinese translation, landed on Xue Yue's desk.   Fifteen days later, at the Changsha Victory Celebration, unit accolades were proclaimed; for "shooting down enemy plane, obtaining vital enemy documents," meritorious honors went to 134th Commander Yang Gancai and Deputy Liu. Each received 1000 yuan and one 3rd Class Baoding Medal. Okamura's October 2 order original: Chinese forces retreated to Miluo and Xiushui Rivers banks assembling; to avoid disadvantage, this army should quickly withdraw to original positions, restore combat strength.   Withdrawal plan as follows: … Xue's October 3 order original:   "Northern Hunan frontal units with current posture immediately pursue facing enemy fiercely, must capture in Chongyang-Yueyang south area. ... Pursuit units may detach part to monitor and sweep enemy collection troops; main force execute overtaking pursuit... Already deep behind enemy advance units vigorously destroy enemy transport lines, cut escape routes."   From October 3, Chinese forces unleashed ferocious counteroffensives against the Japanese on three fronts: northern Hunan, southern Hubei, and the Hunan-Hubei-Jiangxi border; the invaders receded like a vanishing tide, never to reclaim their ground. The 25th and 195th Divisions hounded the 6th Division and Nara Detachment from Fulinpu back to the Miluo River, then to the Xin Qiang River. On October 8, the Japanese fled across the Xin Qiang; the 195th's 566th Brigade surged in pursuit, launching a nocturnal raid on Xitang-Jianshan. Gains were modest, but the enemy, entrenched in their den, resisted with feral tenacity. Qin commanded the brigade's withdrawal southward; northern Hunan operations concluded.   In southern Hubei, the 79th Army chased remnants of the 33rd Division from Sanyan Bridge to Pingjiang, across Nanjiang Bridge, hounding them back to their Tongcheng lair. On the Hunan-Hubei-Jiangxi border, 30th Army Group Commander Wang Lingji orchestrated a pincer against Japanese at Xiushui. The foes retreated to Sandu, mounting a stubborn defense. Chinese assaults faltered for three days; on the fourth night's blitz, victory crowned their efforts, expelling the invaders to their original Wuning stronghold. With both armies reclaiming pre-war lines, the First Battle of Changsha drew to its resounding close.   Over days, Xue Yue received a deluge of congratulatory telegrams and letters from the Nationalist Government, Military Committee, National Assembly, myriad civic groups, party officials, and social luminaries. As hoped, among them was Chiang Kai-shek's effusive missive, brimming with joy. For Xue Yue, this one sufficed. Chiang Kai-shek's telegram to Xue Yue:   "In this northern Hunan campaign, over half the enemy was annihilated. The triumphant news has invigorated the nation, all due to effective command and soldiers' valor; I commend without reservation. Thoroughly investigate and report meritorious personnel from this battle; also report the dead and wounded for awards and relief. With this initial victory foundation laid, our officers and men's responsibilities grow heavier; urge your subordinates to extra vigilance, redoubled effort, avoiding arrogance or complacency, to amass great achievements, my deepest hopes."   As if countering Chongqing's high-powered broadcasts, Japanese radios in Wuhan, Nanjing, Beiping, and Manchukuo blared at full volume: "In this Xiang-Gan operation, valiant Imperial forces penetrated over 100 km into northern Hunan, sweeping anti-peace elements, routing Chinese central main forces, inflicting over 40,000 enemy casualties, a pivotal triumph advancing the holy war. Having achieved objectives, Imperial troops have victoriously withdrawn..."   In the aftermath of the First Battle of Changsha, the Japanese high command spun a tale of calculated restraint, insisting their assault was merely a spoiling raid, a calculated jab never intended to seize and hold the city indefinitely. With brazen confidence, they downplayed their toll, claiming a mere 850 souls lost to death and 2,700 wounded in the fray, while boastfully asserting they had slain 44,000 Chinese defenders and taken 4,000 captive, painting a picture of overwhelming triumph amid the smoke and ruin. Yet, foreign military observers, peering through the fog of propaganda with detached scrutiny, painted a starkly different canvas. They gauged Chinese losses at a far more tempered 20,000 killed and wounded, a heavy but bearable scar on the nation's resolve, while estimating Japanese casualties soared to around 30,000, a grievous hemorrhage that belied the invaders' claims of minimal sacrifice. Military historian Michael Clodfelter, sifting through the annals of conflict, ventured an even grimmer tally: a staggering 50,000 Japanese casualties endured in the relentless clash, a testament to the ferocity of Chinese resistance and the high price of imperial ambition. In the battle's locale, neither side claimed clear victory, but globally for the resistance, it favored China. I would like to take this time to remind you all that this podcast is only made possible through the efforts of Kings and Generals over at Youtube. Please go subscribe to Kings and Generals over at Youtube and to continue helping us produce this content please check out www.patreon.com/kingsandgenerals. If you are still hungry after that, give my personal channel a look over at The Pacific War Channel at Youtube, it would mean a lot to me. The First Battle of Changsha unfolded in September 1939 during China's War of Resistance Against Japan. Japanese forces under Okamura Yasuji advanced into Hunan and Jiangxi, crossing rivers and capturing key positions like Yingtian amid fierce Chinese defenses led by Xue Yue. 

Go to https://shortlinus.com to touch grass. Get a free 15-day trial of Odoo's all-in-one business solution and see how it can make your life easier! Check it out at https://www.odoo.com/wan Visit https://www.squarespace.com/WAN and use offer code WAN for 10% off We're all Born Private, and it should stay that way. Create a free account with Proton Mail https://proton.me/wan Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2 NewGen: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

Ed @ff_litigator and Jeremiah @coachretzlaff1 break down their top best ball buys and fades with the NFL Draft less than four weeks away and the ADP landscape settling in. Buys discussed: Omarion Hampton (RB8, ~Round 2) Jakobi Meyers (WR41, ~Round 7–8) Josh Downs (WR52, Round 10–11 turn) Wan'dale Robinson (Round 10) Jonathon Brooks (RB47, Round 13) Woody Marks (Round 14) Fades discussed: Javonte Williams (RB19, Round 4) Alec Pierce (WR, Round 6) Jaylen Warren — take the discount on Rico Doell instead (Round 8) KC Concepcion (Round 10 price point) Dalton Kincaid (TE10, Round 10) Brenton Strange (Round 13) Learn more about your ad choices. Visit megaphone.fm/adchoices

Join Ryan Wormeli, Andrew Erickson, and Jake Ciely as they highlight their favorite wide receiver targets in each round of 2026 fantasy football drafts! Timestamps: (May be off due to ads) Intro - 0:00:00 Justin Jefferson - 0:05:15 Chris Olave - 0:10:12 Tailgate Podcast - 0:14:04 Rashee Rice - 0:14:40 Davante Adams - 0:19:41 Hard Rock Bet - 0:24:05 Jaylen Waddle - 0:25:37 DJ Moore - 0:33:25 FantasyPros Discord - 0:38:25 Marvin Harrison Jr. - 0:38:53 Ricky Pearsall - 0:43:25 Wan’Dale Robinson - 0:47:08 Jayden Higgins - 0:49:36 Josh Downs - 0:52:56 Romeo Doubs - 0:55:01 Outro - 1:00:03 Helpful Links: Hard Rock Bet - All lines provided by Hard Rock Bet. Hard Rock Bet is giving out $25 Bonus Bets throughout the tournament if a team you bet to win or cover hits a buzzer beater. If you haven’t joined Hard Rock Bet yet, now’s the time to check in the game. New signups can double their winnings on their first ten bets – max fifty dollars. That means if you would’ve won a hundred bucks on your bet, now it’s two hundred. Head over to Hard Rock Bet, sign up and make your first deposit today. Payable in bonus bet(s). Not a cash offer. Offered by the Seminole Tribe of Florida in FL. Offered by Seminole Hard Rock Digital, LLC, in all other states. Must be 21+ and physically present in AZ, CO, FL, IL, IN, MI, NJ, OH, TN or VA to play. Terms and conditions apply. Concerned about gambling? In FL, call 1-833-PLAYWISE. In IN, if you or someone you know has a gambling problem and wants help, call 1-800-9-WITH-IT. GAMBLING PROBLEM? CALL 1-800-GAMBLER (AZ, CO, IL, MI, NJ, OH, TN, VA). Follow us on Twitch - The team here at FantasyPros is taking questions all week, every week on Twitch. Follow us on Twitch at twitch.tv/fantasypros and never miss a stream! Discord – Join our FantasyPros Discord Community! Chat with other fans and get access to exclusive AMAs that wind up on our podcast feed. Come get your questions answered and BE ON THE SHOW at fantasypros.com/chat Leave a Review – If you enjoy our show and find our insight to be valuable, we’d love to hear from you! Your reviews fuel our passion and help us tailor content specifically for YOU. Head to Apple Podcasts, Spotify, or wherever else you get your podcasts and leave an honest review. Let’s make this show the ultimate destination for fantasy football enthusiasts like us. Thank you for watching and for showing your support – https://fantasypros.com/review/ BettingPros Podcast – For advice on the best picks and props across both the NFL and college football each and every week, check out the BettingPros Podcast at bettingpros.com/podcast, our BettingPros YouTube channel at youtube.com/bettingpros, or wherever you listen to podcasts.See omnystudio.com/listener for privacy information.

If you want to skip the hassle of researching, buying, and building a gaming PC for yourself, buy one from one of Jawa's Verified Sellers! Visit https://jawa.link/WANMar26 to get started. Use code: WANSHOW10 to get 10% OFF (up to $100) your first order. Visit https://www.squarespace.com/WAN and use offer code WAN for 10% off Thanks to Tello for sponsoring this video! Check out their plans at https://tello.com/?utm_source=partner&utm_medium=cpc&utm_campaign=WAN&src=partner&mdm=cpc&cmg=WAN We're all Born Private, and it should stay that way. Create a free account with Proton Mail: https://proton.me/wan Get a Circuit Board skin for your device so dbrand can keep messing with Linus at https://dbrand.com/pcb Check out the Razer Blade series of laptops; perfect for work or pleasure: https://lmg.gg/wanrazerblade Game or work in comfort on a Razer Iskur V2: https://lmg.gg/wanrazeriskur Get a special deal on Private Internet Access VPN today at https://www.piavpn.com/LinusWan Purchases made through some store links may provide some compensation to Linus Media Group. Learn more about your ad choices. Visit megaphone.fm/adchoices

Join Joe Pisapia, Derek Brown, and special guest Justin Boone from Yahoo Sports as they reveal their top 10 must-have fantasy players to target in 2026 now that the dust has settled from NFL free agency! Timestamps: (May be off due to ads) Intro - 0:00:00 Isaiah Likely - 0:03:35 Chig Okonkwo - 0:07:15 Jaylen Waddle Trade Live Reactions - 0:11:25 Hard Rock Bet - 0:15:50 Mike Evans - 0:18:10 Wan’Dale Robinson - 0:22:25 Ken Walker - 0:26:11 Chuba Hubbard - 0:30:42 Bhayshul Tuten - 0:33:31 FantasyPros Discord - 0:35:24 Omarion Hampton - 0:36:10 Malik Willis - 0:38:28 Ashton Jeanty - 0:43:07 Outro - 0:45:39 Helpful Links: Hard Rock Bet - All lines provided by Hard Rock Bet. ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Sign up for Hard Rock Bet and make a $5 bet and you'll get $150 in bonus bets if you win. Head over to Hard Rock Bet, sign up and make your first deposit today. Payable in bonus bet(s). Not a cash offer. Offered by the Seminole Tribe of Florida in FL. Offered by Seminole Hard Rock Digital, LLC, in all other states. Must be 21+ and physically present in AZ, CO, FL, IL, IN, MI, NJ, OH, TN or VA to play. Terms and conditions apply. Concerned about gambling? In FL, call 1-888-ADMIT-IT. In IN, if you or someone you know has a gambling problem and wants help, call 1-800-9-WITH-IT. GAMBLING PROBLEM? CALL 1-800-GAMBLER (AZ, CO, IL, MI, NJ, OH, TN, VA). Follow us on Twitch - The team here at FantasyPros is taking questions all week, every week on Twitch. Follow us on Twitch at twitch.tv/fantasypros and never miss a stream! Discord – Join our FantasyPros Discord Community! Chat with other fans and get access to exclusive AMAs that wind up on our podcast feed. Come get your questions answered and BE ON THE SHOW at fantasypros.com/chat Leave a Review – If you enjoy our show and find our insight to be valuable, we’d love to hear from you! Your reviews fuel our passion and help us tailor content specifically for YOU. Head to Apple Podcasts, Spotify, or wherever else you get your podcasts and leave an honest review. Let’s make this show the ultimate destination for fantasy football enthusiasts like us. Thank you for watching and for showing your support – https://fantasypros.com/review/ BettingPros Podcast – For advice on the best picks and props across both the NFL and college football each and every week, check out the BettingPros Podcast at bettingpros.com/podcast, our BettingPros YouTube channel at youtube.com/bettingpros, or wherever you listen to podcasts.See omnystudio.com/listener for privacy information.

The Giants made key Free agent signings on the first day of Free agency. Justin and Shaun react to the signings and give their thoughts on Giants players who have left and where the Giants go next. Start your free online visit today at https://Hims.com/giants for your personalized ED treatment options. Use our code for 10% off your next SeatGeek order*: https://seatgeek.onelink.me/RrnK/JOMBOY10. Sponsored by SeatGeek. *Restrictions apply. Max $20 discount For a limited time Hollow Socks is having a Buy 2, Get 2 Free Sale. Head to Hollowsocks.com today to check it out. #Hollow Sockspod 00:00 Giants Sign Isaiah Likely, Tremaine Edmunds & Jermaine Eluemunor 01:00 First thoughts about the Giants FA signings 05:15 Jermaine Eluemunor Stays in NY 09:00 a huge value signing 11:00 What this means for Marcus Mbow 13:00 Isaiah Likely Follows Harbaugh to the Giants 20:20 One of the best blocking TE's in the league 23:00 A big time receiving threat 27:35 Replacing both Wan'dale and Bellinger 31:15 Tremaine Edmunds is a great culture addition 35:15 Giants will ask their LBs to play coverage 40:00 An experienced leader of the defense 45:06 Hits hard 50:00 Jordan Stout highest paid punter in NFL history 55:50 Ar'Darius Washington a Dane Belton replacement 58:21 re-signing Hodges and Manhertz 01:02:00 Looking for WR signing 01:04:00 BTJ trade possible? 01:07:30 Brian Daboll signs Giants players to the Titans 01:13:00 What do you want picked up next Check out our Merch: https://shop.jomboymedia.com/collections/talkin-giants Subscribe to JM Football for our NFL coverage: https://www.youtube.com/@JMFootball Follow all of our content on https://jomboymedia.com #giants #nygiants Hosted by Simplecast, an AdsWizz company. See pcm.adswizz.com for information about our collection and use of personal data for advertising.

Join Joe Pisapia and Andrew Erickson as they break down all of the major moves from the first day of NFL Free Agency! Timestamps: (May be off due to ads) Intro - 0:00:00 Malik Willis to Miami Dolphins - 0:01:01 Kenneth Walker to Kansas City Chiefs - 0:04:42 Travis Etienne to New Orleans Saints - 0:09:51 Tyler Allgeier to Arizona Cardinals - 0:12:44 Mike Evans to San Francisco 49ers - 0:15:25 Alec Pierce back to Indianapolis Colts and Michael Pittman Jr. to Pittsburgh Steelers - 0:19:28 Wan’Dale Robinson to Tennessee Titans - 0:22:22 Isaiah Likely to New York Giants - 0:24:46 Jalen Nailor and Tyler Linderbaum to Las Vegas Raiders - 0:26:40 Kenneth Gainwell to Tampa Bay Buccaneers - 0:27:25 JK Dobbins back to Denver Broncos - 0:27:30 Tua Tagovailoa to Atlanta Falcons - 0:28:52 Outro - 0:30:55 Helpful Links: Hard Rock Bet - All lines provided by Hard Rock Bet. ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Sign up for Hard Rock Bet and make a $5 bet and you'll get $150 in bonus bets if you win. Head over to Hard Rock Bet, sign up and make your first deposit today. Payable in bonus bet(s). Not a cash offer. Offered by the Seminole Tribe of Florida in FL. Offered by Seminole Hard Rock Digital, LLC, in all other states. Must be 21+ and physically present in AZ, CO, FL, IL, IN, MI, NJ, OH, TN or VA to play. Terms and conditions apply. Concerned about gambling? In FL, call 1-888-ADMIT-IT. In IN, if you or someone you know has a gambling problem and wants help, call 1-800-9-WITH-IT. GAMBLING PROBLEM? CALL 1-800-GAMBLER (AZ, CO, IL, MI, NJ, OH, TN, VA). Follow us on Twitch - The team here at FantasyPros is taking questions all week, every week on Twitch. Follow us on Twitch at twitch.tv/fantasypros and never miss a stream! Discord – Join our FantasyPros Discord Community! Chat with other fans and get access to exclusive AMAs that wind up on our podcast feed. Come get your questions answered and BE ON THE SHOW at fantasypros.com/chat Leave a Review – If you enjoy our show and find our insight to be valuable, we’d love to hear from you! Your reviews fuel our passion and help us tailor content specifically for YOU. Head to Apple Podcasts, Spotify, or wherever else you get your podcasts and leave an honest review. Let’s make this show the ultimate destination for fantasy football enthusiasts like us. Thank you for watching and for showing your support – https://fantasypros.com/review/ BettingPros Podcast – For advice on the best picks and props across both the NFL and college football each and every week, check out the BettingPros Podcast at bettingpros.com/podcast, our BettingPros YouTube channel at youtube.com/bettingpros, or wherever you listen to podcasts.See omnystudio.com/listener for privacy information.

Join Joe Pisapia, Andrew Erickson, and Jake Ciely as they break down the Top 60 WRs in early consensus rankings and highlight players they are higher or lower on than the field! Timestamps: (May be off due to ads) Intro - 0:00:00 WRs 1-12 - 0:00:36 Chris Olave - 0:01:42 Malik Nabers - 0:04:40 Justin Jefferson - 0:06:41 FantasyPros Dynasty Channel - 0:10:02 WRs 13-24 - 0:10:32 DeVonta Smith - 0:13:10 Luther Burden - 0:13:53 Davante Adams - 0:15:17 Zay Flowers - 0:17:18 WRs 25-36 - 0:20:37 Emeka Egbuka - 0:21:12 Brian Thomas - 0:22:42 Rome Odunze - 0:24:25 Marvin Harrison - 0:26:02 WRs 37-48 - 0:27:01 Wan’Dale Robinson - 0:27:21 Quentin Johnston - 0:29:02 Ricky Pearsall - 0:30:18 WRs 49-60 - 0:31:43 Deebo Samuel - 0:32:12 Josh Downs - 0:33:16 Brandon Aiyuk - 0:34:50 Chimere Dike - 0:36:25 Outro - 0:37:26 Helpful Links: Hard Rock Bet - All lines provided by Hard Rock Bet. ⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠⁠Sign up for Hard Rock Bet and make a $5 bet and you'll get $150 in bonus bets if you win. Head over to Hard Rock Bet, sign up and make your first deposit today. Payable in bonus bet(s). Not a cash offer. Offered by the Seminole Tribe of Florida in FL. Offered by Seminole Hard Rock Digital, LLC, in all other states. Must be 21+ and physically present in AZ, CO, FL, IL, IN, MI, NJ, OH, TN or VA to play. Terms and conditions apply. Concerned about gambling? In FL, call 1-888-ADMIT-IT. In IN, if you or someone you know has a gambling problem and wants help, call 1-800-9-WITH-IT. GAMBLING PROBLEM? CALL 1-800-GAMBLER (AZ, CO, IL, MI, NJ, OH, TN, VA). Follow us on Twitch - The team here at FantasyPros is taking questions all week, every week on Twitch. Follow us on Twitch at twitch.tv/fantasypros and never miss a stream! Discord – Join our FantasyPros Discord Community! Chat with other fans and get access to exclusive AMAs that wind up on our podcast feed. Come get your questions answered and BE ON THE SHOW at fantasypros.com/chat Leave a Review – If you enjoy our show and find our insight to be valuable, we’d love to hear from you! Your reviews fuel our passion and help us tailor content specifically for YOU. Head to Apple Podcasts, Spotify, or wherever else you get your podcasts and leave an honest review. Let’s make this show the ultimate destination for fantasy football enthusiasts like us. Thank you for watching and for showing your support – https://fantasypros.com/review/ BettingPros Podcast – For advice on the best picks and props across both the NFL and college football each and every week, check out the BettingPros Podcast at bettingpros.com/podcast, our BettingPros YouTube channel at youtube.com/bettingpros, or wherever you listen to podcasts.See omnystudio.com/listener for privacy information.