😈 WebGym is now 300% faster with better performance. We introduce AsyncWebRL, with 300% end-to-end throughput and better performance over WebGym (which uses REINFORCE) with async GRPO. Two algorithmic takeaways we learnt: (1) do asynchronous multi-step agentic RL with the **decoupled loss**, and (2) **don’t do normalization over step-number**, instead use the same weighting over each step across trajectories. Details below 🧵: Paper: arxiv.org/pdf/2606.05597 Website: asyncwebrl-website.github.io… Code: github.com/microsoft/webgym 1/11

Introducing CUA-Lite 🧵 — an open platform for computer-use agents. Training and benchmarking CUAs (Computer-Use Agents) requires four core pieces: 1️⃣ Agents — the models and the scaffolding that drives them 2️⃣ Environments — runtime/sandboxes for agents to interact with, tasks & verifiers/graders 3️⃣ Traces — records of agent trajectories 4️⃣ Frameworks — to evaluate, SFT & RL-train agents Today, all four are fragmented. Every agent ships with its own implementation, often in a separate repo — there is no unified way to run them all. Every environment exposes its own interface and action space, often requiring an expensive VM sandbox for each verifiable task. Traces come in incompatible formats. And without common standards across the stack, every project ends up rebuilding its own tooling/framework for eval, SFT, and RL. CUA-Lite unifies the stack: → One standardized interface & action space for agents and environments → One standardized format for agent traces → One framework for evaluation, SFT & RL → Across desktop, browser & mobile And open resources plug straight in, creating the largest open collection of CUA agents, environments and traces, all in a unified format: 🤖 10+ CUAs, including GPT, Claude, Gemini, Qwen, Muse-Glimmer, UI-TARS 🌐 15+ benchmarks, including OSWorld, WebArena & AndroidWorld ⚡ Optional VM-free sandboxes with 30K+ verifiable tasks for training 📚 10+ trace datasets, freely available on Hugging Face, including public datasets converted into the standardized format and fresh rollouts from frontier open-weight CUAs Led by @BerkeleyRDI , our goal is for CUA-Lite to become a community-driven, open-source ecosystem for computer-use agents. Join the community and contribute today: bring an environment (runtime/sandbox + tasks + verifier), traces, or an agent, and plug it into CUA-Lite!

Behavioral cloning mystery seohong.me/blog/behavioral-c… I wrote a new blog post about "mysteries" in behavioral cloning that appear with real-world robot data (e.g., overfitting is "good"). I also tried to demystify them and shared my thoughts!

on side note, the fun of working in RL theory is that results like this are still discovered from time to time, where a few lines explain (to experts) significance + proof machinery, and we are like "how come we didn't realize this before"

So glad to see that WebGym is being used as an environment in training Qwen-AgentWorld!!

Check out Rui's paper on a fully open framework for training visual web agents with online multi-turn RL on live websites, where a tiny 4B model hits open-source SOTA and even rivals proprietary systems!

Tomorrow at ExHall A & F Poster Location: 471, 5pm-7pm, we'll present WebGym, together introducing two recently pre-released works (AsyncWebRL and OpenWebRL) as a surprise :) WebGym: webgym-website.github.io/web… OpenWebRL: openwebrl.github.io AsyncWebRL: asyncwebrl-website.github.io

I will be attending CVPR in Denver on June 5 and June 6, presenting our work WebGym, the largest yet visual web agentic RL framework, along with many interesting observations we found during RL. Come by if you're interested! Track the event at cvpr.thecvf.com/virtual/2026… :)

Today we launch Recursive. We are building AI that discovers knowledge automatically and improves itself recursively, an open-ended process that will fundamentally change how science and technology advance. Our 25 top researchers and engineers in San Francisco and London bring diverse expertise spanning agentic AI scientists, architecture and algorithm design, world models, optimization, and interpretability, united by a shared conviction that this is the most important problem we could be working on today. If you are interested in joining, please send your resume to talent@recursive.com. Follow us at @Recursive_SI!

Congrats, Dr. Tong!