Model Releases
TimeWarp: Evaluating Web Agents by Revisiting the Past
arXiv:2603.04949v2 Announce Type: replace-cross Abstract: As web agents close the gap with humans on benchmarks, one question arises: Do today's agents perform just as well on tomorrow's web? We intro
arXiv:2603.04949v2 Announce Type: replace-cross Abstract: As web agents close the gap with humans on benchmarks, one question arises: Do today's agents perform just as well on tomorrow's web? We introduce TimeWarp, a benchmark that emulates the evolving web. TimeWarp consists of three web environments, each with six UI versions spanning UI design, frontend code, and workflows from different eras of the internet. We pair TimeWarp with a set of complex, realistic tasks covering different forms of web navigation. Our experiments reveal that vision-based agents are vulnerable to changes, while text-based agents become brittle once fine-tuned on a single version. To address this, we propose TimeTraj, a new annotation method that uses plan distillation to collect trajectories across multiple versions. By training agents on teacher rollouts using our BC-variant, we achieve substantial performance gains: 20.4% to 37.7% for Qwen-3 4B and 0% to 27.0% for Llama-3.1 8B models. Our work helps study generalization across web designs and opens a new paradigm for collecting plans rather than trajectories to improve the robustness of web agents.
Related
- Do Maps Still Matter for Machines: Revisiting the Role of Choropleth Maps in Foundation Model Spatial Understanding
- SecureWebArena: A Holistic Security Evaluation Benchmark for LVLM-based Web Agents
- Benchmarking General Mobile Assistants in Challenging Real-World Scenarios
Source: arXiv cs.CV | 2026-09-09