model releases

arXiv:2607.04171v1 Announce Type: cross Abstract: Large Vision-Language Models (LVLMs) have shown strong multimodal understanding and spatial grounding, but their computational cost limits real-time r

128 GB of memory is nice, but you can get started with local agentic AI workflows with much less. By connecting gemma 4 in @lmstudio to MATLAB MCP Server, you can run a local AI model that uses MATLAB

This post discusses how observing Claude's internal activation patterns (J-space) reveals the model's hidden reasoning processes, such as detecting code bugs and analyzing images, even when these step

Claude Fable running ComfyUI workflows through the Comfy MCP is a cheat code. → Claude pulled the shot from the web, auto-detected the scene cuts + trimmed it → ran my saved depthanything v3 + openpos

Armin Ronacher / Armin Ronacher's Thoughts and Writings: Claude Opus 4.8 and Sonnet 5 seem worse at tool calls than older models, likely due to post-training that assumes Claude Code-like harnesses as

During a Bloomberg interview, Yann LeCun (@ylecun ) explains why LLMs are limited in terms of real-world intelligence during a Bloomberg interview. 'Language is a very approximate, reduced, quantized,

'Ghost memory' is a real problem with agents. You might have seen the issue where a long-running agent still confidently repeats a user fact that stopped being true weeks ago? New research names the f

Interesting stuff. And the visualization at the end is worth trying: https://www.neuronpedia.org/qwen3.6-27b/jlens New Anthropic research: A global workspace in language models. Of everything happenin

Less than 2 years later, with Fable: 'simulate an encounter between a mind flayer and a drow warrior of equal CR. Set up initial stats then simulate each move, including initiative, rolling dice as ne

Must-read research by Anthropic. Here is the simple explanation and why this is a big deal. We suspect LLMs perform 'internal reasoning'. But little is known or do good methods exist to understand it.

Ethan Mollick discusses the configuration scope differences between OpenAI and Claude AI systems, distinguishing between global settings that apply across all uses, local machine-specific settings, an

One of the only times I remind people I have a PhD in computational neuroscience is when people without a neuroscience background say their model works 'like the brain.' In these cases, I put on my ne

SpaceX's 17th Transporter rideshare mission is scheduled to launch from California, carrying 81 payloads to orbit. Transporter missions are SpaceX's dedicated rideshare service that launches multiple

Reporting benchmark results as a single scalar percentage is insufficient for meaningful evaluation of model performance. Comprehensive benchmark reporting should include efficiency metrics alongside

The automotive industry is at a pivotal crossroads as it hits the gas on adopting new technology. The era of the traditional connected vehicle has shifted into the age of the software-defined vehicle

Reuters: SK Hynix launches a US share sale on the Nasdaq to raise ~28B, selling 17.79M new shares; the final listing price is due Thursday before trading starts Friday — South Korean chipmaker SK Hyni

so much for recursive self improvement, to the degree that it requires scientific taste the other thing im noticing while working on my research projects is how limited these models are GPT-5.5-xhigh

Release: sqlite-utils 4.0rc3 I hoped to release sqlite-utils 4.0 stable this weekend, but as I worked through the backlog of issues and PRs with a combination of Claude Fable 5 and GPT-5.5 the changel

In this post, you learn how to use the new MLflow integration with Amazon SageMaker AI optimized inference recommendation jobs and Amazon SageMaker AI benchmark jobs to automatically stream experiment

tencent/Hy3 New Apache 2.0 licensed model from Tencent in China: Hy3 is a 295B-parameter Mixture-of-Experts (MoE) model with 21B active parameters and 3.8B MTP layer parameters, developed by the Tence

The Nemotron model family from NVIDIA has reached 100 million downloads, marking a significant milestone for the open-source AI model community. The achievement reflects growing adoption and collabora

This is our first time telling the story of how we first built and launched Claude Code, starting with its origins in Anthropic safety research. So much more to do. We are 1% done. We've put together

“This study cuts through the optimism surrounding medical AI by showing how easily benchmark success can be mistaken for real readiness. In medical AI, impressive scores are clearly not the same as tr

We released our in-house Japanese, English, and Chinese translation tool, Sakana Translate! Try it→ https://translate.sakana.ai 🐟 • Translate: Handles long text in real time • Proofread: Tone and phra

When it comes to Chess, @viditchess is an absolute legend. Recently, he entered the Hermes Agent Accelerated Business Hackathon to see who could build the best agent using Nemotron. Here's what he sub

With our internal coding benchmark, we're able to confidently introduce open-weight models into our AI code reviewer w/o degrading code quality. Have the frontier model (Fable) to the hardest work, de

This post from Anthropic's X account shares a Wordle game result, indicating the player solved Wordle puzzle #1,842 in 3 attempts. The emoji grid shows the color-coded feedback from each guess, with g

As you are hiring AI people, look for folks who have kept up with the trajectory of the industry, especially OSS. Someone who has their ear to the ground. Things are moving FAST. (deepagents existed ~

Wency Chen / South China Morning Post: ByteDance's Doubao and Alibaba's Qwen will disable humanlike and user-created agents before July 15, as China's anthropomorphic AI interaction rules take effect

(deepagents existed ~10 months before EVE, but...) yes - the agent industry has shifted from: ~agent frameworks~ (langchain, ai sdk, llama index) to ~agent harnesses~ (deepagents, claude agent SDK, EV

Fable prompting tips, based on all the guides Anthropic and its employees have given us. Three things you might want to do: 1) review your claude.md file against prompting best practices so you don't

SpaceX's Falcon 9 rocket successfully launched 29 Starlink satellites and BesxarFoundry's Flight 1 payload from Florida. The mission demonstrates continued expansion of the Starlink constellation whil

If you want to keep up with the industry, just follow what the Langchain team are doing it is all open source I’ve been following them since Langgraph multi agent collaboration, to deepagents with san

'Group project, but make it 1776.' That's how a new commercial for Google Workspace opens. And things only get cringier from there. The clip imagines what it would be like if the founding fathers turn

Yohei Nakajima used Claude AI to construct a comprehensive family tree that connected all family members for a weekend gathering. The post highlights a practical application of AI in organizing and vi

Sam Altman expressed amazement at his older child combining two words for the first time, drawing a humorous comparison to being similarly impressed by a hypothetical GPT-5.6 discovery. The post refle

Somewhat humbling to have Claude Fable do a final review of some software that you're about to release and have it then find (and fix) FIVE release blockers, for an estimated (unsubsidized) cost of $1

I wrote about the sqlite-utils 4.0rc1 release a couple of weeks ago. Since we only have Claude Fable on our Max subscriptions for a few more days, I decided to see if it could help me get to a 4.0 sta

“That was fun, I’ll do it again when I get out,” said the 22-year-old Gambian violent offender after he stabbed a 55-year-old Italian man twenty times. And he will be released again. Perhaps to succee

There is a massive demand for processing files *in the agent loop*. The number of users submitting agent queries with file attachments is exponentially increasing over time. Our mission is to make Lit

“What has really happened is VC built the wrong AI, you’ll pay a high price for this error with your pension, and nobody will go to jail.” By far the shrewdest and most entertaining analyst of the AI

This post documents a Wordle game result (puzzle #1,841) solved in 4 attempts, showing the letter position feedback for each guess using the standard color-coded system (gray for incorrect letters, ye

As America turns 250, we put together 250 open AI milestones from the US: open models, datasets, demos, papers, and tools that helped shape the field. They go from attention is all you need, pytorch,

Better Models: Worse Tools Armin reports on a weird problem he ran into while hacking on Pi: The short version is that newer Claude models sometimes call Pi’s edit tool with extra, invented fields in

Hey, Fable: 'No like 'AAA' in what everyone online thinks AAA is, you know what I mean.' This was pretty funny, there are lootboxes, EULAs, achievements, useless graphical settings, elaborate boot scr

i often think about the irony of how 'tools for thought' people spent like a decade making cool pretty demos with canvases and then got completely mogged by low contrast poorly designed CLIs just winn