WebGym: Scaling Training Environments for Visual Web Agents with Realistic Tasks
This work addresses the challenge of non-stationary and diverse real websites for visual web agents, offering a scalable training environment, though it is incremental in improving existing methods.
The authors tackled the problem of training robust visual web agents by introducing WebGym, a large-scale open-source environment with nearly 300,000 tasks across real-world websites, and achieved a success rate improvement from 26.2% to 42.9% on an out-of-distribution test set, outperforming proprietary models like GPT-4o and GPT-5-Thinking.
We present WebGym, the largest-to-date open-source environment for training realistic visual web agents. Real websites are non-stationary and diverse, making artificial or small-scale task sets insufficient for robust policy learning. WebGym contains nearly 300,000 tasks with rubric-based evaluations across diverse, real-world websites and difficulty levels. We train agents with a simple reinforcement learning (RL) recipe, which trains on the agent's own interaction traces (rollouts), using task rewards as feedback to guide learning. To enable scaling RL, we speed up sampling of trajectories in WebGym by developing a high-throughput asynchronous rollout system, designed specifically for web agents. Our system achieves a 4-5x rollout speedup compared to naive implementations. Second, we scale the task set breadth, depth, and size, which results in continued performance improvement. Fine-tuning a strong base vision-language model, Qwen-3-VL-8B-Instruct, on WebGym results in an improvement in success rate on an out-of-distribution test set from 26.2% to 42.9%, significantly outperforming agents based on proprietary models such as GPT-4o and GPT-5-Thinking that achieve 27.1% and 29.8%, respectively. This improvement is substantial because our test set consists only of tasks on websites never seen during training, unlike many other prior works on training visual web agents.