Search: “hardware”

Yes, open-source / open-weight models are important for a healthy AI ecosystem. That's how we can verify things, check claims, and keep up outside the closed labs. Plus, it gives us the freedom to run

My 2hr workshop on Open vs Closed models, reward hacking, benchmaxxing & RL is out! 1. Closed vs open models 2. Throughput maxxing but accuracy minimizing 3. Benchmaxxing & cheating 4. Distillation &

75.4% SWE Bench Verified / 53.9% SWE Bench Pro on 1 bit quantisation is 🤪 This is in line with my expectations & you can expect even lower drop off with NVP4 base trained models - why not run everythi

New model for AMD Strix Halo users: My 198B Step 3.7 Flash release was a big hit, but this one may be even better: 298B-parameter Hy3, now running on a new 2-bit FPX codebook designed to map efficient

It's incredible to me that 9 months ago we didnt have an office, and now we have a fully functional humanoid and an efficient AI! So proud of the team 🤠 Starting with the fundamentals Prototype Versio

reminds me of the time Greg Brockman personally downloaded YouTube videos for training purposes, as reported in NYT. 🦔Apple is suing OpenAI for systematic trade secret theft weeks before the IPO. The

Ollama announced support for running multiple open-source language models locally, emphasizing accessibility and ease of deployment for users who want to use AI models without relying on cloud service

The upcoming wave of SpaceXAI Grok updates is insane Grok 4.5: The 1.5T foundation model is being refined almost daily, and its context window is expected to jump to 1M tokens, possibly as soon as nex

- OpenAI continually outperforms Anthropic models on computer use (with Claude, I’m surprised when it works, with Codex, I expect it to work) - when I prompt with Fable 5.5, I feel like I’m motivating

A study from @Stanford showed that 71.3% of chatgpt queries could be accurately answered by a local model. I suspect a major part of enterprise AI workloads could be run locally too for free (compared

Earlier this year, @SpaceX acquired @xAI (now SpaceXAI), which operates the Colossus datacenters in Memphis. As SpaceX continues to invest in the area, SpaceX is offering our neighbors in the Memphis

Half price Starlink for people in the Memphis region SpaceXAI has announced that they will be applying a 50% discount on the standard @Starlink monthly price for both new and existing customers in the

Today, we give robots a /skills library that self-evolves and compounds indefinitely! Introducing ASPIRE: a robot solving its 100th task is no longer as clueless as solving its first. Coding agents ob

MASSIVE NEWS has partnered with NVIDIA to advance local AI deployment as a default option, addressing the shift toward on-device and edge AI processing rather than cloud-dependent models. This collabo

Sam Altman announced that OpenAI's inference speed will reach 750 tokens per second on GPUs with 5.6 TFLOPS in July. This represents a significant increase in throughput capability for OpenAI's models

In HF GGUF section of models, we are emphasizing MTP heads with its own sign 𝗠𝗧𝗣 llama.cpp adds MTP for the Qwen3.6 family This is a significant milestone for the local AI ecosystem. The performance j

Elon Musk built one of the largest AI compute clusters on earth. Yann LeCun just explained why xAI now rents it out to rivals instead of winning with it. Musk has antagonized so much AI talent he stru

HF is quietly becoming the best place to store data, public AND private, especially for brutal domains like robotics and video AI where the files are massive, append-only, and never stop growing. Exam

I have some very big news... KernelBench-Hard with H100 and B200 (single gpu results) AND KernelBench-Mega tested on RTX PRO 6000, H100, B200 is finally out! Starting with Mega, each of models wrote a

Emad Mostaque predicts that open-source AI models will soon reach a capability level ('fable-level') where they can run efficiently on consumer-grade MacBook Air/Mini computers without specialized har

Qualcomm is a piece of shit supplier to anyone trying to innovate on the edge, especially startups. My client is trying to buying ~500,000 of SoCs and wants them to be American because patriotism. Ins

anyone who wants to put money into OpenAI’s IPO should think again. the talent is voting with its feet, and has been for several years. Personal update: I’ve decided to leave OpenAI. I’m proud to have

And another open-weight release. Nemotron 3 Ultra has an ultra impressive capability:efficiency ratio! Design-wise, it carries forward the Mamba-2-attention hybrid stack and LatentMoE introduced in th

Highlighting recent advances in multi-GPU and tensor parallel support in llama.cpp Over the last few months llama.cpp maintainers and engineers from NVIDIA collaborated to improve the multi-GPU perfor

Many research labs only consider inference efficiency after the fact. Step 3.7 Flash is a 196B MoE model, and built for inference from the start by @StepFun_ai. Multi-Matrix Factorization Attention (M

This post from Nous Research references coordinates (likely geographic) and announces a new era in PC computing. The content likely discusses recent developments in personal computing technology, poss

a hot (cold at this point?) take that lead us to build this: every agent in the future will need a sandbox to connect to writing/executing code is not just for coding agents! is useful for all sorts o

AI infrastructure is shifting toward vertical integration, where companies build specialized, end-to-end solutions optimized for specific use cases rather than relying on generic horizontal platforms.

Behind the MiMo API Price Reduction: The deepest price cut, up to 99%, is for Input (Cache Hit). The core reason is our inference framework now supports hierarchical KV cache optimization for SWA. Pro

Today, Zyphra Research is sharing fundamental work extending Equilibrium Propagation beyond Energy-Based Models to biologically realistic neuron models. A step toward more efficient AI, local learning

Command A+ is available on @huggingface with W4A4 quantization 🤗 Cut your serving footprint dramatically with virtually zero performance degradation. Try it now: https://huggingface.co/CohereLabs/comm

Spent Tuesday watching @Google I/O thinking about a question. If you build on @Android, what just changed for you? Short answer: your screen belongs to Gemini now. After this week, Google owns the AI,

We built a bipedal robot for about $2,500. A real, mostly 3D-printed robot you can build, repair, simulate, train, and control. Today we’re releasing LeRobot Humanoid: an open robot-learning platform

It's been *almost* a bit quiet around LLM architecture releases in the past two weeks 😅 Interesting tidbit is the parallel block design. Via the Cmd-A the tech report 'equivalent performance but signi

Elon Musk commented positively on an all-black paint finish applied to a base or foundation of an unspecified product or structure. The post suggests aesthetic appreciation for the sleek appearance cr

wait… did Cohere just release Command A+ models under Apache 2.0 for the first time ever?! 🙊 welcome to Europe! 🤗 Introducing: Cohere Command A+ We’ve created our most powerful LLM yet, optimized it t

Google I/O is a virtual event featuring two keynote presentations: a Consumer Keynote at 10:00 AM PT and a Developer Keynote at 1:30 PM PT. The event showcases Google's latest products, innovations, a

llama.cpp adds MTP for the Qwen3.6 family This is a significant milestone for the local AI ecosystem. The performance jump with these changes is massive and elevates local inference on commodity hardw

llama.cpp with MTP support makes local models fast enough to use as daily drivers 🚀 Qwen3.6-27B dense generation (on A10G): From 25 tok/s → 45 tok/s (+78%). Two flags on llama-server: --spec-type draf

This playbook provides step-by-step instructions for running Nous Research's Hermes Agent locally on NVIDIA DGX Spark using Ollama, enabling users to deploy an open-source AI agent entirely on local h

NVIDIA's RTX AI Garage partnership with Nous Research demonstrates deploying Hermes agent models on DGX systems integrated with Apache Spark for accelerated AI workloads. The initiative showcases how

Clem Delangue announced the Reachy Mini robot is operational with audio capabilities enabled, with plans to integrate it with local AI services and a Hermes Agent in the near future. This indicates pr

Reachy Mini, a wireless robotic platform, has been launched in Dubai, with Clement Delangue promoting its potential for developing agentic AI systems. The announcement highlights the availability of t

Researchers developed a control method for Reachy Mini robots that uses a real human face as an input controller rather than building a humanoid facial interface. This approach leverages existing robo

We started running monthly research sessions at the @agi_inc office. First one was last Friday on PNM compilation. If you do on-device research and want to come present, apply here: https://form.typef

Hello again, everyone! Our latest Qwopus3.6-35B-A3B-v1 is now live, and it is once again breathtaking! Full HF space benchmark showcase and write-up is in the comments, so you can make conclusions for