Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

This is why local models must win. Cloud models refused to reverse-engineer Apple's RDMA protocol. DeepSeek and Nemotron said, "hold my beer". TBF, Nemotron needed KV Cache injection to comply lol.

Ash Hart

2,251 subscribers

59,901 views • 7 days ago •via X (Twitter)

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

NVIDIA just released a new open source transcription model, Nemotron Speech ASR, designed from the ground up for low-latency use cases like voice agents. Here's a voice agent built with this new model. 24ms transcription finalization and total voice-to-voice inference time under 500ms. This agent actually uses *three* NVIDIA open source models: - Nemotron Speech ASR - Nemotron 3 Nano 30GB in a 4-bit quant (released in December) - A preview checkpoint of the upcoming Magpie text-to-speech model These models are all truly open source: weights, training data, training code, and inference code. This is a big deal! Jensen said in the CES keynote yesterday that he expects open source models to catch up to proprietary models this year in a number of categories. NVIDIA is putting their weight behind making this happen. (As Alan Kay said, the best way to predict the future is to invent it.) The code for this agent is open source too, of course. You can deploy it to production with Modal and Pipecat AI cloud, or run locally on an NVIDIA DGX Spark or RTX 5090.

NVIDIA just released a new open source transcription model, Nemotron Speech ASR, designed from the ground up for low-latency use cases like voice agents. Here's a voice agent built with this new model. 24ms transcription finalization and total voice-to-voice inference time under 500ms. This agent actually uses three NVIDIA open source models: - Nemotron Speech ASR - Nemotron 3 Nano 30GB in a 4-bit quant (released in December) - A preview checkpoint of the upcoming Magpie text-to-speech model These models are all truly open source: weights, training data, training code, and inference code. This is a big deal! Jensen said in the CES keynote yesterday that he expects open source models to catch up to proprietary models this year in a number of categories. NVIDIA is putting their weight behind making this happen. (As Alan Kay said, the best way to predict the future is to invent it.) The code for this agent is open source too, of course. You can deploy it to production with Modal and Pipecat AI cloud, or run locally on an NVIDIA DGX Spark or RTX 5090.

kwindla

274,679 views • 7 months ago

Perplexity Computer will soon be able to dynamically split compute power between local models and cloud models! If that would drive Perplexity Computer costs down, it would be huge, since it is one of the top blockers for many at this moment. Soon 👀

Perplexity Computer will soon be able to dynamically split compute power between local models and cloud models! If that would drive Perplexity Computer costs down, it would be huge, since it is one of the top blockers for many at this moment. Soon 👀

🚨 AI News | TestingCatalog

26,787 views • 2 months ago

Open models are accelerating development and deployment across agentic AI, physical AI, and a growing range of industries. ✨ “We have billions of dollars of supercomputers in operation, so we can develop our open models,” said NVIDIA CEO Jensen Huang in a recent keynote. “That work is starting to attract attention across industries worldwide, as we build frontier AI models in many different domains.” This includes leading open models from NVIDIA, including: ⚙️ NVIDIA Nemotron for agentic AI ⚙️ NVIDIA Cosmos and GR00T for physical AI ⚙️ NVIDIA Alpamayo for autonomous vehicles ⚙️ NVIDIA Clara for biomedical AI Learn more about open source and AI ➡️

Open models are accelerating development and deployment across agentic AI, physical AI, and a growing range of industries. ✨ “We have billions of dollars of supercomputers in operation, so we can develop our open models,” said NVIDIA CEO Jensen Huang in a recent keynote. “That work is starting to attract attention across industries worldwide, as we build frontier AI models in many different domains.” This includes leading open models from NVIDIA, including: ⚙️ NVIDIA Nemotron for agentic AI ⚙️ NVIDIA Cosmos and GR00T for physical AI ⚙️ NVIDIA Alpamayo for autonomous vehicles ⚙️ NVIDIA Clara for biomedical AI Learn more about open source and AI ➡️

NVIDIA AI

19,296 views • 7 months ago

nvidia's 3B mamba destroyed alibaba's 3B deltanet on the same RTX 3090. only 24 days between releases. same active parameters, same VRAM tier, completely different architectures. nemotron cascade 2: 187 tok/s. flat from 4K to 625K context. zero speed loss. flags: -ngl 99 -np 1. that's it. no context flags, no KV cache tricks. auto-allocates 625K. qwen 3.5 35B-A3B: 112 tok/s. flat from 4K to 262K context. zero speed loss. flags: -ngl 99 -np 1 -c 262144 --cache-type-k q8_0 --cache-type-v q8_0. needed KV cache quantization to fit 262K. both models held a flat line across every context level. both architectures are context-independent. but nvidia's mamba2 is 67% faster at generating tokens on the exact same hardware and needs fewer flags to get there. same node, same GPU, same everything. the only variable is the model. gold medal math olympiad winner running at 187 tokens per second on single RTX 3090 a card from 6 years ago. nvidia cooked.

nvidia's 3B mamba destroyed alibaba's 3B deltanet on the same RTX 3090. only 24 days between releases. same active parameters, same VRAM tier, completely different architectures. nemotron cascade 2: 187 tok/s. flat from 4K to 625K context. zero speed loss. flags: -ngl 99 -np 1. that's it. no context flags, no KV cache tricks. auto-allocates 625K. qwen 3.5 35B-A3B: 112 tok/s. flat from 4K to 262K context. zero speed loss. flags: -ngl 99 -np 1 -c 262144 --cache-type-k q8_0 --cache-type-v q8_0. needed KV cache quantization to fit 262K. both models held a flat line across every context level. both architectures are context-independent. but nvidia's mamba2 is 67% faster at generating tokens on the exact same hardware and needs fewer flags to get there. same node, same GPU, same everything. the only variable is the model. gold medal math olympiad winner running at 187 tokens per second on single RTX 3090 a card from 6 years ago. nvidia cooked.

Sudo su

187,008 views • 4 months ago

QVAC SDK 0.10.0 is now live, bringing advanced local compute capabilities and specialized hardware optimization directly to your device Key Features and Updates: - Image-to-Image Diffusion: Transform and edit images using simple prompts with 100% local compute—no cloud uploads or external servers required - Dynamic Tooling & KV Cache Management:Your local LLM now receives a tailored toolbox for every interaction, with automatic KV cache clearing to maintain high-speed inference - Doctor CLI: A new diagnostic tool that analyzes your hardware and memory, providing specific instructions on how to optimize your GPU for local AI - Suspend & Resume API: Specifically designed for mobile environments, this allows apps to pause P2P swarms and RAG workspaces to meet background rules without losing model state - GPT-OSS Compatibility: Added support for the latest GPT-OSS models loaded externally, expanding the range of open-source intelligence available on the platform Build the future of private, unstoppable AI:

QVAC SDK 0.10.0 is now live, bringing advanced local compute capabilities and specialized hardware optimization directly to your device Key Features and Updates: - Image-to-Image Diffusion: Transform and edit images using simple prompts with 100% local compute—no cloud uploads or external servers required - Dynamic Tooling & KV Cache Management:Your local LLM now receives a tailored toolbox for every interaction, with automatic KV cache clearing to maintain high-speed inference - Doctor CLI: A new diagnostic tool that analyzes your hardware and memory, providing specific instructions on how to optimize your GPU for local AI - Suspend & Resume API: Specifically designed for mobile environments, this allows apps to pause P2P swarms and RAG workspaces to meet background rules without losing model state - GPT-OSS Compatibility: Added support for the latest GPT-OSS models loaded externally, expanding the range of open-source intelligence available on the platform Build the future of private, unstoppable AI:

QVAC

34,508 views • 3 months ago

.Cohere is defining the future of sovereign AI. Autumn Moulder, VP of Engineering at Cohere, breaks down the requirements for a full AI sovereign stack at #NVIDIAGTC: ✅ Full-Stack Sovereignty: Deploying everything from the model to applications and reasoning traces within a single data center. ✅ The Power of Openness: Why open models like NVIDIA Nemotron are critical for data lineage and navigating evolving regulations.

.Cohere is defining the future of sovereign AI. Autumn Moulder, VP of Engineering at Cohere, breaks down the requirements for a full AI sovereign stack at #NVIDIAGTC: ✅ Full-Stack Sovereignty: Deploying everything from the model to applications and reasoning traces within a single data center. ✅ The Power of Openness: Why open models like NVIDIA Nemotron are critical for data lineage and navigating evolving regulations.

NVIDIA AI

10,965 views • 4 months ago

Guangdong's AI+ Chronicles | The world's most unique testing ground for AI innovation isn't Silicon Valley. Watch on to discover why!👇 DeepSeek's rise has spurred industries to deploy open-source models, driving a major industrial shift. Partnerships like "#DeepSeek + Huawei Ascend Cloud" and "DeepSeek + Tencent" are leading the way. But in this wave, #Guangdong stands apart. Rather than focusing solely on large models and computing power, the province prioritizes application-driven innovation, ensuring #AI tangibly transforms everyday life. What makes Guangdong so special? 🧐

Guangdong's AI+ Chronicles | The world's most unique testing ground for AI innovation isn't Silicon Valley. Watch on to discover why!👇 DeepSeek's rise has spurred industries to deploy open-source models, driving a major industrial shift. Partnerships like "#DeepSeek + Huawei Ascend Cloud" and "DeepSeek + Tencent" are leading the way. But in this wave, #Guangdong stands apart. Rather than focusing solely on large models and computing power, the province prioritizes application-driven innovation, ensuring #AI tangibly transforms everyday life. What makes Guangdong so special? 🧐

Glocal

102,541 views • 1 year ago

That time they accused Meghan of breaking protocol for going bare feet on the beach , and Prince Harry said hold my beer while I take a stroll with my wife bare feet skipping in to the sunset 🤭 He is a such Diana son .

That time they accused Meghan of breaking protocol for going bare feet on the beach , and Prince Harry said hold my beer while I take a stroll with my wife bare feet skipping in to the sunset 🤭 He is a such Diana son .

Claire

165,250 views • 1 year ago

holy sh*t this is f**king insane I cancelled my $200/mo Claude for this I replaced opus 5 with deepseek v4-flash and someone figured out to how to use them as subagents on codex [it takes 3 mins to set up here is how] 1. install this repo called 'model-router' 2. toggle subagent models > 'all selected models' 3. that's it

holy sh*t this is f**king insane I cancelled my $200/mo Claude for this I replaced opus 5 with deepseek v4-flash and someone figured out to how to use them as subagents on codex [it takes 3 mins to set up here is how] 1. install this repo called 'model-router' 2. toggle subagent models > 'all selected models' 3. that's it

Avid

415,791 views • 14 days ago

We are opening up the beta of our new AI writing tool to Exclusive Models. If you're exclusive to JFF, try it out and let us know your thoughts. At our Anniversary Conference, lots of models said they needed help making their posts more creative. This is a tool to help models take what they write and help them make better posts. Does this replace a human? No! In fact, if you are an erotic writer, apply to be in our Business Directory so models can hire you to help them out!

Sensitive content

We are opening up the beta of our new AI writing tool to Exclusive Models. If you're exclusive to JFF, try it out and let us know your thoughts. At our Anniversary Conference, lots of models said they needed help making their posts more creative. This is a tool to help models take what they write and help them make better posts. Does this replace a human? No! In fact, if you are an erotic writer, apply to be in our Business Directory so models can hire you to help them out!

JustForFans

15,325 views • 3 years ago

local LLMs on Mac are getting good. but the workflow around models is still messy. you have models in: •⁠ ⁠LM Studio •⁠ ⁠Ollama •⁠ ⁠Hugging Face cache •⁠ ⁠random folders •⁠ ⁠manual downloads and every runtime expects something slightly different. we built ModelHub to clean up this layer. it’s a native macOS menu bar app for discovering, downloading, and managing local LLMs. not a replacement for Ollama, oMLX or LM Studio. it sits beside them. Ollama = run models MLX = Apple Silicon acceleration LM Studio = local model workspace Hugging Face = model ecosystem ModelHub = model discovery + management layer we’re live on Product Hunt today. would love feedback from people running local models on Mac.

local LLMs on Mac are getting good. but the workflow around models is still messy. you have models in: •⁠ ⁠LM Studio •⁠ ⁠Ollama •⁠ ⁠Hugging Face cache •⁠ ⁠random folders •⁠ ⁠manual downloads and every runtime expects something slightly different. we built ModelHub to clean up this layer. it’s a native macOS menu bar app for discovering, downloading, and managing local LLMs. not a replacement for Ollama, oMLX or LM Studio. it sits beside them. Ollama = run models MLX = Apple Silicon acceleration LM Studio = local model workspace Hugging Face = model ecosystem ModelHub = model discovery + management layer we’re live on Product Hunt today. would love feedback from people running local models on Mac.

sabesh 📟

31,497 views • 2 months ago

Inside Nemotron and NVIDIA's AI lab: my conversation with Bryan Catanzaro (Bryan Catanzaro). NVIDIA is a chip company. So why does it put hundreds of researchers on building AI models - and then give them away for free? We go deep into the Nemotron models, what it takes to build a top AI lab, and the future of frontier AI. 01:33 - Is open source AI catching the frontier? 05:29 - Do closed labs blocking distillation slow open source down? 07:42 - Is the US falling behind China? 10:30 - Why companies actually choose open models 12:39 - A "crazy" 2008 bet: machine learning on GPUs 15:33 - Working with Andrew Ng and Dario Amodei at Baidu 17:41 - Coming back to NVIDIA: DLSS and the birth of Megatron 21:55 - The real reason NVIDIA builds its own models 24:28 - Is Moore's Law really dead? 33:37 - The Nemotron family: Nano, Super, Ultra 35:09 - Built for agents: why NVIDIA bets on speed 36:02 - How you train a 550B model in 4 bits 39:25 - Hybrid Mamba-Transformer, explained simply 42:31 - Mixture of experts, and why NVIDIA built NVL72 around it 47:26 - Why a 1-million-token context window matters 49:26 - Multi-token prediction: how the model predicts 5 tokens at once 52:47 - Multi-teacher distillation: teaching one model from many 58:01 - Where reinforcement learning goes next 01:00:16 - Inside NVIDIA's research org: "the mission is the boss" 01:04:03 - How NVIDIA decides who gets the GPUs 01:10:53 - Why NVIDIA still feels entrepreneurial after 33 years 01:12:58 - Why Bryan doesn't believe in the singularity 01:17:50 - The AI backlash 01:19:18 - The controversial case: open AI is safer than closed

Inside Nemotron and NVIDIA's AI lab: my conversation with Bryan Catanzaro (Bryan Catanzaro). NVIDIA is a chip company. So why does it put hundreds of researchers on building AI models - and then give them away for free? We go deep into the Nemotron models, what it takes to build a top AI lab, and the future of frontier AI. 01:33 - Is open source AI catching the frontier? 05:29 - Do closed labs blocking distillation slow open source down? 07:42 - Is the US falling behind China? 10:30 - Why companies actually choose open models 12:39 - A "crazy" 2008 bet: machine learning on GPUs 15:33 - Working with Andrew Ng and Dario Amodei at Baidu 17:41 - Coming back to NVIDIA: DLSS and the birth of Megatron 21:55 - The real reason NVIDIA builds its own models 24:28 - Is Moore's Law really dead? 33:37 - The Nemotron family: Nano, Super, Ultra 35:09 - Built for agents: why NVIDIA bets on speed 36:02 - How you train a 550B model in 4 bits 39:25 - Hybrid Mamba-Transformer, explained simply 42:31 - Mixture of experts, and why NVIDIA built NVL72 around it 47:26 - Why a 1-million-token context window matters 49:26 - Multi-token prediction: how the model predicts 5 tokens at once 52:47 - Multi-teacher distillation: teaching one model from many 58:01 - Where reinforcement learning goes next 01:00:16 - Inside NVIDIA's research org: "the mission is the boss" 01:04:03 - How NVIDIA decides who gets the GPUs 01:10:53 - Why NVIDIA still feels entrepreneurial after 33 years 01:12:58 - Why Bryan doesn't believe in the singularity 01:17:50 - The AI backlash 01:19:18 - The controversial case: open AI is safer than closed

Matt Turck

56,749 views • 1 month ago

NVIDIA is giving away free access to 130+ AI models for a full year > most people building AI agents are paying $50-200/month for API access NVIDIA just made that argument irrelevant models you get: MiniMax M2.7, GLM 5.1, Kimi 2.5, DeepSeek-v4-flash, GPT-OSS-120B and 110+ more setup: > step 1 - get your free key > go to > register -> bind phone -> copy API key > step 2 - add to Hermes agent > open Settings -> Model Provider -> Custom base_url = " api_key = "nvapi-xxxxxxxxxxxxxxxxxxxx" > step 3 - pick a model model = "minimaxai/minimax-m2.7" model = "zhipuai/glm-5.1" model = "moonshot-ai/kimi-2.5" model = "deepseek-ai/deepseek-v4-flash" model = "nvidia/nemotron-3-ultra-550b-a55b" > Hermes already has NVIDIA set as default base_url > paste the key and you're running instantly > works the same in Cursor and OpenCode > cost: $0 > limit: 40 req/min > expires: 1 year while everyone is paying for API access, this is sitting there for free

NVIDIA is giving away free access to 130+ AI models for a full year > most people building AI agents are paying $50-200/month for API access NVIDIA just made that argument irrelevant models you get: MiniMax M2.7, GLM 5.1, Kimi 2.5, DeepSeek-v4-flash, GPT-OSS-120B and 110+ more setup: > step 1 - get your free key > go to > register -> bind phone -> copy API key > step 2 - add to Hermes agent > open Settings -> Model Provider -> Custom base_url = " api_key = "nvapi-xxxxxxxxxxxxxxxxxxxx" > step 3 - pick a model model = "minimaxai/minimax-m2.7" model = "zhipuai/glm-5.1" model = "moonshot-ai/kimi-2.5" model = "deepseek-ai/deepseek-v4-flash" model = "nvidia/nemotron-3-ultra-550b-a55b" > Hermes already has NVIDIA set as default base_url > paste the key and you're running instantly > works the same in Cursor and OpenCode > cost: $0 > limit: 40 req/min > expires: 1 year while everyone is paying for API access, this is sitting there for free

Mr. Buzzoni

195,973 views • 2 months ago

Vik and Val Bercovici (WEKA) map where AI inference memory is headed. Every 100x cut in KV cache gets swallowed by ~10,000x more usage, so demand climbs. - NVLink beats the board: 128 lanes vs 32 PCIe - WEKA serves NAND-backed storage faster than DRAM over network - DeepSeek's cache reads run ~87x cheaper, China only - CXL needs a dedicated bus, WEKA pools NAND over RDMA/NVLink instead - Val's call: SaaS giants and Neoclouds must merge Chapters: 0:00 Intro and memory prices 3:00 Model routing and offloading 5:10 Network faster than motherboard 13:06 Four-tier memory hierarchy 19:40 KV cache and Jevons paradox 23:50 DeepSeek cache read pricing 31:40 Sliding window attention 34:49 NAND tiers, SLC vs QLC 43:57 High bandwidth flash use cases 49:59 CXL versus NVLink 53:44 AMD Mex and small models 1:03:14 SaaS, Neoclouds and tokenomics Get more of Austin and Vik daily, free! Sign up: Connect with Vik and Austin: Vik's Paid Substack: Austin's Paid Substack: Austin Lyons Vikram Sekar

Vik and Val Bercovici (WEKA) map where AI inference memory is headed. Every 100x cut in KV cache gets swallowed by ~10,000x more usage, so demand climbs. - NVLink beats the board: 128 lanes vs 32 PCIe - WEKA serves NAND-backed storage faster than DRAM over network - DeepSeek's cache reads run ~87x cheaper, China only - CXL needs a dedicated bus, WEKA pools NAND over RDMA/NVLink instead - Val's call: SaaS giants and Neoclouds must merge Chapters: 0:00 Intro and memory prices 3:00 Model routing and offloading 5:10 Network faster than motherboard 13:06 Four-tier memory hierarchy 19:40 KV cache and Jevons paradox 23:50 DeepSeek cache read pricing 31:40 Sliding window attention 34:49 NAND tiers, SLC vs QLC 43:57 High bandwidth flash use cases 49:59 CXL versus NVLink 53:44 AMD Mex and small models 1:03:14 SaaS, Neoclouds and tokenomics Get more of Austin and Vik daily, free! Sign up: Connect with Vik and Austin: Vik's Paid Substack: Austin's Paid Substack: Austin Lyons Vikram Sekar

Semi Doped

60,918 views • 1 month ago

Today we have an exclusive. After 7 years at Kleiner Perkins, Bucky Moore is announcing his next chapter. And he’s doing it on 20VC. We go deep on: 💼 Why Mega Platforms Will Win the Future of Venture 📈 Why Every Fund Has to do Pre-Seed To Win Today 🔮 Are Models Venture Investments 🧭 Which Verticals Will Models Want to Own vs Not My 6 key takeaways 👇

Today we have an exclusive. After 7 years at Kleiner Perkins, Bucky Moore is announcing his next chapter. And he’s doing it on 20VC. We go deep on: 💼 Why Mega Platforms Will Win the Future of Venture 📈 Why Every Fund Has to do Pre-Seed To Win Today 🔮 Are Models Venture Investments 🧭 Which Verticals Will Models Want to Own vs Not My 6 key takeaways 👇

Harry Stebbings

28,527 views • 1 year ago

José on why AI infrastructure is winning this cycle. “You can imagine that the cloud business will be commoditized eventually, but right now it’s actually really hard to serve these models… And so there’s just margin there.”

José on why AI infrastructure is winning this cycle. “You can imagine that the cloud business will be commoditized eventually, but right now it’s actually really hard to serve these models… And so there’s just margin there.”

Delphi Digital

27,729 views • 21 days ago

Today we're announcing that hybrid agentic inference is coming to Perplexity Computer. Computer can split tasks between a local model running on your machine and frontier models in the cloud. This keeps private data on your device and maximizes token efficiency. Coming soon.

Today we're announcing that hybrid agentic inference is coming to Perplexity Computer. Computer can split tasks between a local model running on your machine and frontier models in the cloud. This keeps private data on your device and maximizes token efficiency. Coming soon.

Perplexity

353,429 views • 2 months ago

DeepSeek R1 671B is ridiculously good, but running at 158 tokens per second is something out of this world. A model *this good* running *this fast* is 3 years ahead of this time. The secret: DeepSeek in this short video isn't running on GPUs. It's running on custom-made chips! This is running on SambaNova and their SN40L chips, which are designed and optimized for running AI workflows. SambaNova is sponsoring my work. We designed GPUs to run games and later realized we could use them for AI. This is very different from creating a chip specifically for AI from the ground up. Just think about this for a second: One single SNL40 chip can simultaneously hold 100+ models (trillions of parameters) in memory! An agentic workflow that uses multiple models simultaneously won't need to swap models from memory even once! By the way, the Llama family of models *screams* on the SambaNova Cloud. Here are some numbers from my latest tests: • Llama 3.3 70B: 367.32 t/s • Llama 3.2 1B: 2381.82 t/s • Llama 3.2 3B: 1335.71 t/s • Llama 3.1 8B: 929.58 t/s • Llama 3.1 70B: 369.16 t/s • Llama 3.1 405B: 103.37 t/s You can access all of these models for free!

DeepSeek R1 671B is ridiculously good, but running at 158 tokens per second is something out of this world. A model this good running this fast is 3 years ahead of this time. The secret: DeepSeek in this short video isn't running on GPUs. It's running on custom-made chips! This is running on SambaNova and their SN40L chips, which are designed and optimized for running AI workflows. SambaNova is sponsoring my work. We designed GPUs to run games and later realized we could use them for AI. This is very different from creating a chip specifically for AI from the ground up. Just think about this for a second: One single SNL40 chip can simultaneously hold 100+ models (trillions of parameters) in memory! An agentic workflow that uses multiple models simultaneously won't need to swap models from memory even once! By the way, the Llama family of models screams on the SambaNova Cloud. Here are some numbers from my latest tests: • Llama 3.3 70B: 367.32 t/s • Llama 3.2 1B: 2381.82 t/s • Llama 3.2 3B: 1335.71 t/s • Llama 3.1 8B: 929.58 t/s • Llama 3.1 70B: 369.16 t/s • Llama 3.1 405B: 103.37 t/s You can access all of these models for free!

Santiago

131,019 views • 1 year ago