Search: “models”

Jeremy Howard argues that AI model developers at OpenAI and Anthropic should prioritize breaking backwards compatibility during the prototyping phase rather than maintaining it, suggesting that backwa

NEW paper worth reading. A full agentic workflow can be distilled into model weights and run at roughly 100x lower inference cost while preserving near-frontier task quality. The workflow includes mul

Managed Deep Agents is a new LangChain feature in private beta that provides managed infrastructure for deep agents, designed to be model-agnostic and deployable with minimal code. The service aims to

Cohere is on such a great open-source trajectory lately. Beautiful Apache 2.0 model! https://huggingface.co/CohereLabs/command-a-plus-05-2026-bf16 Command A+ from @cohere is out now :) its our best mo

It’s a really special time to be alive…some thoughts from training this model 🧵 Today, we share a breakthrough on the planar unit distance problem, a famous open question first posed by Paul Erdős in

Running the same CNA search on base models (before instruction tuning) yields a structurally similar set of neurons, but ablating them produces almost no behavioral change. We read this as evidence th

This post explores conceptual parallels between classical distributed systems design patterns—such as actor models and reactive blackboard architectures—and their potential application to LLM agent de

⚡️After weeks of hard work, we're thrilled to fully open-source HY World 2.0 today -- full inference code and all models! Build, explore, and create your own interactive worlds with us.👇 https://githu

Introducing Composer 2.5, our most powerful model yet. It's more intelligent, better at sustained work on long-running tasks, and more reliable at following complex instructions. For the next week, we

Large language models demonstrate a specific and persistent theory-of-mind failure despite excelling at other theory-of-mind tasks, suggesting a particular gap in their reasoning about beliefs, intent

Note that as of last summer Waymo said that LLMs/VLMs were experimental, and that their foundation model then could “process only a small amount of image frames, does not incorporate accurate 3D sensi

Together with SpaceXAI, we’re training a significantly larger model from scratch, using 10x more total compute. With Colossus 2’s million H100-equivalents and our combined data and training techniques

The best feature of @xai Grok Build right now is how it handles subagents and personas. Most people still treat the model like one very smart intern that has to do everything at once. Grok Build took

Interesting interpretability paper on tool-using agents. The authors probe hidden states and find the model often recognizes it should call a tool, but fails to actually call one. The mismatch ranges

New Template Batch Drop SELFIE modeled by @paranoidream Classic Bedroom Mirror https://grok.com/imagine/templates/8a017403-bbad-428c-8900-5e338c7f3eb4 Post-Shower Fresh Look https://grok.com/imagine/t

Ollama 0.24 now supports integration with a Codex app, allowing users to run open-source models through the application after updating to the latest version. The feature enables users to select and ut

DID SOMEONE SAY FREE TOKENS IN FLEET???? Yes it's true. Fleet now has a built in model powered by @FireworksAI_HQ that's free for all Developer & Plus plan users. Try it out today (or before @hwchase1

LangSmith Fleet has introduced a free model option powered by Fireworks AI for users on Developer and Plus plans, lowering the barrier to entry for getting started with the platform. This update makes

you can see more here (color stuff is at bottom): http://colorfeelings.replit.app - changing embedding model may change results - the similarity range was pretty small so normalizing really stretched

Our evaluations show that frontier AI's cyber capabilities are advancing quickly. The length of cyber tasks frontier models can complete has been doubling every few months, and this rate has become fa

i like that there are models called bert and ernie, but in all seriousness, this update looks impressive ERNIE 5.1 is here 🚀 ERNIE 5.1 significantly reduces pretraining cost while compressing total pa

In KAME, a fast speech model starts replying instantly, while a backend LLM runs in parallel to inject deep knowledge on the fly. It’s a completely different way to approach conversational AI, making

Talked to a friend at a top AI lab. Their whole team is former journalists, training models on poems, summaries, and creative writing. I use AI every day and can see that it tends to flatten my writin

Ethan Mollick discusses a tradeoff in OpenAI's newer realtime voice model, where improved capabilities require developers to revise prompts that were written for the previous version. The post highlig

This post from Nous Research discusses the integration of computer use capabilities with Hermes Agent models in collaboration with Claude (CUA), enabling AI agents to interact with computer interfaces

For browser-use AI agents, every task is dozens of model calls in a tight loop. The inference layer isn’t background infrastructure. It’s what the product runs on. @yutori_ai runs Scouts, Delegate, an

In modern ML accelerators, FLOPS have absolutely exploded. Often though, the bottleneck is not FLOPS but memory bandwidth. Similarly, model intelligence has exploded, causing the bottleneck to be huma

real time model streaming audio/video/text in and out with tool use People talk, listen, watch, think, and collaborate at the same time, in real time. We've designed an AI that works with people the s

The inability of AI models to produce creative variation is a huge gap. The fact that they generate similar ideas limits their ability to do science & the same-y writing limits their usefulness in man

Frontier labs are betting AGI models will be so good you won't ever want to customize them. We think different. Building on a closed platform means renting your intelligence. The landlord sets the ter

This post from Harrison Chase discusses how AI agents have evolved beyond just the underlying model, emphasizing critical infrastructure components including state management, observability, retry mec

Sam Altman solicited feedback from the public on X (formerly Twitter) regarding desired improvements for OpenAI's next model release. The post likely gathered community input on priorities such as rea

This post draws a parallel between Python's design philosophy (from 'The Zen of Python') and machine learning model capabilities, suggesting that if a skill or implementation can be easily explained,

Jeremy Howard proposes three key enhancements for large language models: incorporating latent space diffusion-like reasoning processes, adding a persistent recurrent state mechanism, and implementing

This post discusses practical frameworks for working with AI models, covering how context serves as infrastructure, taste functions as configuration, verification enables autonomous systems, and deleg

Sam Altman suggests that OpenAI's competitive advantage extends beyond just offering competitive API rate limits; the superior quality of their models themselves is what keeps users engaged long-term.

Sam Altman solicited examples from users of innovative applications and capabilities enabled by GPT-5.5 that were not feasible with earlier OpenAI models, with particular interest in specific use case

Ethan Mollick advocates for dedicated funding toward developing independent benchmarking systems for AI models, noting that evaluation is currently predominantly conducted by the labs that create the

i thought on device ai was stupid but now my local voice model turns 4x faster on my corporate m4 max laptop than my m2 max personal laptop now i want to have a beefier computer to run the transcripti

Sakana AI Labs has developed Fugu, a multi-agent orchestration system designed to function as a foundation model. The system likely enables coordinated interaction between multiple AI agents to improv

Sam Altman reflects on the trade-off between model performance improvements and practical considerations like cost and speed, suggesting that increased intelligence remains a primary focus despite use

You don't have to choose between either. It's best to use a combination of them. My advice is to learn how to use a few of these models in different harnesses. Learn to combine their strengths. Open-w

i havent done the work to compare it to peers but i'm just excited that we have a base model and honestly for all the people that complained about the death of the completions API (@deepfates ? or dee

Elon Musk allegedly admitted during legal testimony that xAI is distilling OpenAI models to train its own systems and utilizing OpenAI's technology, according to a post by Clem Delangue discussing sta

Our agent harness makes models inside Cursor faster, smarter, and more token-efficient. Here's how we test improvements to the harness, monitor and repair degradations, and customize it for different

LangChain is offering multiple harness variants that are custom-fitted to specific use cases and models, recognizing that effective harness implementation requires tailored configuration rather than o