Gemma 3 is out! We are focused on bringing you open models with best capabilities while being fast and easy to deploy: - 27B lands an ELO of 1338, all the while still fitting on 1 single H100! - vision support to process mixed image/video/text content - extended context window

Today we are releasing the best open-weights model you can run on a single device reaching 1339 Elo on LMsys for Gemma 3 27B (aka zizou-10)! Very strong capabilities on math, multilingual, coding, instruction following, function calling !

Wow!!! Congratulations!!!

Founders who were PhD or post-doc in my lab at Berkeley, **largely funded by NSF / DoD grants**, start-up, market cap (collected by OpenAI Deep Research)

Support arXiv.org during Cornell Giving Day on March 13, 2025. We at Hugging Face are happy to match donations! 💟💟

Come support arXiv.org during Cornell Giving Day on March 13, 2025. Hugging Face is matching donations!

The open source Conversational Speech Model is out. We shared the 1B base model for everyone to build on. Links below.

Match ALREADY completed?! 🤯 Thank you to Hugging Face for this generous match & THANK YOU to everyone who has donated to arXiv on #GivingDay so far!🥳👏🏻🤗 3 MORE HOURS left to give & support #openscience! givingday.cornell.edu/campaigns/arxiv (also stay tuned for a super secret mission)

LLM-based review systems lack domain expertise and structured evaluation. This results in superficial feedback and inaccurate assessments. This paper introduces DeepReview, a multi-stage framework mimicking expert revietheyrs. It uses structured analysis and evidence-based

🚀🌐Introducing AgentRxiv: a framework where autonomous research agents can upload, retrieve, and build on each other’s research. AgentRxiv takes your research direction and progressively outputs research, building on its previous work with each new paper! 🧵

🎉Read the preprint: agentrxiv.github.io Try out AgentRxiv: github.com/SamuelSchmidga… Let’s explore how agents can accelerate research—together. 🧵8/8

🚀 Introducing 🧭MAGELLAN—our new metacognitive framework for LLM agents! It predicts its own learning progress (LP) in vast natural language goal spaces, enabling efficient exploration of complex domains.🌍✨Learn more: 🔗 arxiv.org/abs/2502.07709 #OpenEndedLearning #LLM #RL

Available now in the MCP Marketplace: Browser Use

Gemini 2.5 Pro, available to all users in Cursor! It’s time to build 🚢

Imagine AI doing science: reading papers, generating ideas, designing and running experiments, analyzing results… How many more discoveries can we reveal? 🧐 Meet CodeScientist, a promising next step toward autonomous scientific discovery. 🧵

Today, we're excited to make two big announcements! - OpenHands LM: The strongest 32B coding agent model, resolving 37.4% of issues on SWE-bench Verified 📈 - OpenHands Cloud: SOTA open-source coding agents from your computer, phone, github, with $50 in free credits 🙌☁️

I'm thrilled to share our first release as the AGI SF Lab. Meet Nova Act -- the most effortless way to build agents that can reliably use browsers, giving agents access to much of our digital world. It brings us closer to building universal agents in both digital and physical

📣 This just in... The *open source* and official GitHub MCP Server is here 🤩 github.blog/changelog/2025…

Introducing The AI Scientist-v2, which produced the 1st fully AI-generated paper to pass peer review at a workshop level (at ICLR 2026 2025)! Tech Report: pub.sakana.ai/ai-scientist-v… GitHub: github.com/SakanaAI/AI-Sc… V2 adds: 🧠 Agentic Tree Search (open-ended idea exploration) 💡

Delighted to share the full pre-print and 100% open source code for The AI Scientist-v2 🚀🚀 Tons of improvements enabling modern foundation models to come up with better ideas, conduct deeper and mor general experiments and write better papers! 👩‍🔬