Fara-7B is our first agentic small language model for computer use. This experimental model includes robust safety measures to aid responsible deployment. Despite its size, Fara-7B holds its own against larger, more resource-intensive agentic systems: https://t.co/lT9m0uH4gQ

Studying a paper: Continuous Thought Machines ( what a name! ) fascinating paper from @SakanaAILabs introducing a new neural network architecture that tries to bring back the temporal dynamics of biological brains into AI

This paper shocked me 🤯 Everyone on X keeps bragging about “LLM-as-a-judge” like it’s some magical truth oracle. But this paper shows something insane: Most LLM evaluations you’ve seen are biased by design not because models are bad, but because the judge itself quietly

🚨 This project just made most AI agents look outdated. It’s called Better Agents and it supercharges your coding assistant (Kilocode, Claude Code, Cursor, etc), making it an expert in any agent framework you choose (Agno, Mastra, etc) and all their best practices. This is the

New Microsoft + Oxford + National Univ of Singapore paper shows how computer programs can help redesign websites so other programs use them more reliably. They also build a benchmark of many small web apps and tasks for testing this whole setup. Regular sites are tuned for

This paper shows how a simple memory system called BREW makes language based agents smarter and more efficient. Standard language agents forget what they learned, so every task feels new and they waste many steps repeating trial and error. BREW fixes this by turning past task

working on this model, with this team, with all of the infrastructure we built to get here, is the most rewarding thing i've ever been a part of still can't believe i get to wake up every day and work on this stuff with these people and put all the code for free on the internet

New paper from Qwen team. Gives a gentler way to train reasoning models so they learn more stably. In reinforcement learning fine tuning, a language model generates several answers, receives a reward for each, and updates its weights. Existing group methods such as Group

🚀 Qwen3-VL Tech report is now out on arXiv! From pretraining to post-training, architecture to infra, data to evaluation — we’ve packed in the details for anyone building on vision-language models. 🔥 3 models >1M downloads in just over a month 🏆 Qwen3-VL-8B leads with 2M+

🏆 We are incredibly honored to announce that our paper, "Gated Attention for Large Language Models: Non-linearity, Sparsity, and Attention-Sink-Free" has received the NeurIPS 2025 Best Paper Award! A huge congratulations to our dedicated research team for pushing the boundaries

DSPy Bengaluru - Quarterly Meetup in December. Register here - https://t.co/fGvCFFvgmH #dspy #meetup Agenda: ​DSPy Crash course - Signatures, Modules, Optimizers. - 45 Minutes ​Building Equity Research Reports - DSPy + Agentic Code - 60 Minutes

This one is pretty nasty - it tricks Antigravity into stealing AWS credentials from a .env file (working around .gitignore restrictions using cat) and then leaks them to a webhooks debugging site that's included in the Antigravity browser agent's default allow-list

If Jensen truly believed AGI was near, Nvidia wouldn't sell a single GPU. If Google truly believed AGI was near, they wouldn't sell TPUs to Meta and Anthropic. Just saying.

This paper argues that AI consciousness and human extinction risk are mostly separate things, even if people often confuse them. Existential risk here means a future AI system powerful enough to end humanity while pursuing its own goals. The author says this danger mainly

The paper shows how to replace most manual prompt writing with tiny natural language tags attached directly to code. These tags, called SemTexts, boost Meaning Typed Programming accuracy by about 3x on hard apps while keeping prompt related code roughly 4x smaller. Manual

A large part of humanity lives automated lives; I'm not judging whether that's good or bad, but from this scenario it's possible to completely recreate these people through programming.

New paper from China labs builds a big benchmark and an agent system that turn research papers into clean, editable architecture diagrams. Making these diagrams by hand is slow and subjective, and general image generators ignore the exact logic, structure, and text they should

Potential drug discovery in one 10 min session using Claude Scientific Skills: 265 papers → 14 targets → 8 novel molecules → 1 lead compound (LYM-002) for lymphedema. First molecule designed specifically to enhance lymphatic pump function. Optimized for topical delivery. Try

A very useful tool for integrating autonomous optimization pipelines with @DSPyOSS GEPA. https://t.co/kQAgNEYaLc

> So you could imagine that the deployment itself will involve some kind of a learning trial and error period. It's a process, as opposed to, you drop the finished thing.