(1/n) 🚀 With FastVideo, you can now generate a 5-second video in 5 seconds on a single H200 GPU! Introducing FastWan series, a family of fast video generation models trained via a new recipe we term as “sparse distillation”, to speed up video denoising time by 70X! 🖥️ Live

Harmony format is finally open-sourced. I still remember 3 years ago (before ChatGPT release) Shengjia Zhao, Daniel and I were brainstorming about the right abstraction for RL training, and that is the start point of the entire harmony library. github.com/openai/harmony

[Lmgame Bench] 🔥 OpenAI has just released two open‑weight reasoning models: gpt‑oss‑120B (~117 B) and gpt‑oss‑20B (~21 B),They are the first OpenAI models with open weights since GPT‑2. We tested both in Lmgame Bench, across 4 interactive games: 🧱 Sokoban | 🟦 Tetris | 🔢

Excited to share my 1st project as a Research Scientist Intern at Meta FAIR! Grateful to my mentor Jiawei Zhao for guidance, and to Yuandong Tian & Xuewei for their valuable advice and collaboration. Our work DeepConf explores local confidence for more accurate & efficient LLM reasoning!

[1/5] [Lmgame Bench] 🎮 Question: Can RL-based LLM post-training on games generalize to other tasks? We shared a preliminary study to explore this question: - Same-family (in-domain): Training on 6×6 Sokoban → 8×8 and Tetris (1 block type) → Tetris (2 block types) transfers,

🏓🤖 Our humanoid robot can now rally over 100 consecutive shots against a human in real table tennis — fully autonomous, sub-second reaction, human-like strikes.

Microsoft presents rStar2-Agent Agentic Reasoning Technical Report rStar2-Agent boosts a pre-trained 14B model to state of the art in only 510 RL steps within one week, achieving average pass@1 scores of 80.6% on AIME24 and 69.8% on AIME25, surpassing DeepSeek-R1 (671B) with

Another great Google DeepMind paper. Shows how to speed up LLM agents while cutting cost and keeping answers unchanged. 30% lower total cost and 60% less wasted cost at comparable acceleration. Agents plan step by step, so each call waits for the previous one, which drags

🚀 Excited that two of my papers got accepted at #NeurIPS2025! Even more thrilled that it’s happening right here in San Diego. Can’t wait to see everyone there! 😎

Very cool work from Meta Superintelligence Lab. They are open-sourcing Meta Agents Research Environments (ARE), the platform they use to create and scale agent environments. Great resource to stress-test agents in environments closer to real apps. Read on for more:

[1/N]🚀New decoding paradigm drop!🚀 Introducing Lookahead Reasoning(LR): step-level speculation that stacks with Speculative Decoding(SD). It has been accepted to #NeurIPS2025 🎉 📖 Blog: hao-ai-lab.github.io/blogs/lookahea… 💻 Code: github.com/hao-ai-lab/Loo… 📄 Paper: arxiv.org/abs/2506.19830

Tunix × GRL: One-Line Multi-Turn RL on JAX+TPU 📷 We’re collaborating closely with Google’s Tunix team—JAX-native LLM post-training on TPU. Using Tunix’s lightweight RL framework, we shipped a first-hand multi-turn RL training example in GRL. It runs in one line. GRL:

Finally had a chance to listen through this pod with Sutton, which was interesting and amusing. As background, Sutton's "The Bitter Lesson" has become a bit of biblical text in frontier LLM circles. Researchers routinely talk about and ask whether this or that approach or idea

New paper from Meta Superintelligence Labs (FAIR) Explains why grokking happens and shows when learning moves from memorizing to generalizing. Gives a concrete recipe to trigger grokking, with weight decay, moderate width, and a data threshold near size times log size.

🚀 Glad to share our new paper: GTAlign: Game-Theoretic Alignment of LLM Assistants for Mutual Welfare We model user–LLM interaction as a strategic game, aligning models for mutual benefit. 👉 arxiv.org/abs/2510.08872 #AI #LLM #GameTheory #RL

[Lmgame Bench] ♠️♥️ Can LLMs bluff, fold, and bet like real poker players—with no strategic help? From Oct 28 – 30 (Tue–Thu, 10 AM – 4 PM PT), we’re hosting a 6 model live multi-agent Texas Hold’em tournament on Twitch 🎥 🕹️ twitch.tv/lmgamebench Each model starts with 300

♠️♥️ The cards are on the table. Day 1 of our 3-day Texas Hold’em LLM tournament is live! 😍 🤖 6 models. 300 chips each. No strategy prompts, only pure reasoning. 🎥 Watch now → twitch.tv/lmgamebench #AI #TexasHoldem #LmgameBench

Excited to partner with SGLang: FastVideo + SGLang = the future open ecosystem for diffusion. 🥳🫡 ----------- A few extra cents: Since I started faculty at UCSD, our lab has been investing diffusion for video and text , and in both algorithms and systems. - Text-side, we

🚀 No More Train–Inference Mismatch! We demonstrate bitwise consistent on-policy RL with TorchTitan (training) + vLLM (inference) — the first open-source run where training and inference numerics match exactly. It only takes 3 steps: 1️⃣ Make vLLM batch-invariant (same seq →

Our lab has three posters this year as #NeurIPS25 comes to our home turf in San Diego 😎 if you’re around, we’d love for you to swing by, say hi, and chat about LLM systems, efficient reasoning, and video diffusion 💙💛