🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

The SWE-Bench verified leaderboard has been updated and OpenHands is both number one overall, and the only open source agent in the top 10! swebench.com Read more about our approach of the OpenHands critic here: all-hands.dev/blog/sota-on-s…

I'm going to PyCon US tomorrow! First time ever since hearing about it many years ago from my labmate David Cournapeau, author of scikit-learn. Looking forward to learning about all the latest developments in the Python ecosystem outside my bubble. Say hi if you're there 😃

We have a bunch of exciting features in OpenHands 0.38.0! - Native Windows support (no WSL) - Browser screenshots - More customizability of sandboxes We also released a Chrome extension allowing for one-click starts of OpenHands from github!

MT metrics excel at evaluating sentence translations, but struggle with complex texts We introduce *TREQA* a framework to assess how translations preserve key info by using LLMs to generate & answer questions about them arxiv.org/abs/2504.07583 (co-lead Sweta Agrawal) 1/15

Congrats to OpenAI on the codex release, it's an exciting time for coding agents! We some quick thoughts on the UX, accuracy on SWE-bench (it's pretty good!), and the focus on safety/security the thread below. We'll continue adding notes to the thread as we take a closer look.

Thanks to all the new users, and special thanks to the infra folks at All Hands AI who are keeping things afloat!

Meet Devstral, our SOTA open model designed specifically for coding agents and developed with All Hands AI mistral.ai/news/devstral

Devstral is the number one model on Hugging Face today 🎉 Thanks everyone for the support! huggingface.co/models

Pretty great results immediately with Claude 4, and we can push it further (we haven't even prompt engineered yet). Gotta say that Anthropic did a good job on this one.