InternLM/WildClawBench

Registry

Summary

WildClawBench is an in-the-wild benchmark designed to evaluate AI agents operating within the OpenClaw environment, providing a realistic and challenging testbed for agentic AI systems.

Key features

• In-the-wild benchmark for AI agents
• Built on the OpenClaw environment
• Focuses on agentic AI evaluation
• Realistic and challenging test scenarios

Use cases

• Evaluating AI agent performance in open environments
• Benchmarking agentic AI models
• Research on agentic evaluation methodologies

README excerpt

<h1 align="center">WildClawBench</h1> <p align="center"> <img src="assets/lobster_battle.png" alt="WildClawBench Lobster" width="480"> </p> <div align="center"> []() []() []() [](https://internlm.github.io/WildClawBench/) <br> [](https://arxiv.org/abs/2605.10912) [](https://huggingface.co/papers/2605.10912) [](https://huggingface.co/datasets/internlm/WildClawBench) [](https://github.com/InternLM/WildClawBench/blob/main/WildClawBench_report.pdf) </div> > **Hard, practical, end-to-end evaluation for AI agents — in the wild.** --- **WildClawBench** is an agent benchmark that tests what actually matters: can an AI agent do real work, end-to-end, without hand-holding? We drop agents into a live [OpenClaw](https://github.com/openclaw/openclaw) environment — the same open-source personal AI assistant that real users rely on daily — and throw **60 original tasks** at them: clipping goal highlights from a football match, negotiating meeting times over multi-round emails, hunting down contradictions in search results, writing inference scripts for undocumented codebases, catching privacy leaks before they happen. Useful things. Hard things. Hard enough that **the strongest frontier model we tested still tops out around 62% overall** (technical report Main

Topics