What it measures
- Web navigation
- Multi-step task completion
- Tool and browser control
WebArena
A web-agent benchmark for completing tasks in realistic self-hosted web environments.
How to read the score
% success; higher is better.
Sourced Ranking
| Rank | Model | Provider | Score | Source | Source Type | Verified |
|---|
Limitations
- Environment version and setup can affect comparability.
- The paper is static; project and leaderboard sources are better for current results.
- Record self-hosted environment version when storing scores.
leaderboard
WebArena-x and original WebArena project
The paper is static; project and leaderboard sources are better for current results. Record self-hosted environment version when storing scores.
paper
WebArena paper or specification
Use for benchmark definition, not necessarily for latest model scores.
third-party
WebArena dataset or code
Use for dataset, code, or implementation details; score freshness depends on the benchmark.
third-party
Steel WebArena tracked leaderboard
Third-party tracked leaderboard for WebArena agent setups. Use as a candidate score source only after verifying environment, agent scaffold, and submission details.