Benchmark
BrowserGym WebArena
Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages.
All agentic benchmarksScores from BrowserGym Leaderboard.
- Models measured
- 7
- Scores recorded
- 7
- Sources
- 1
- IQ weight
- 1.08
Score distribution
- 29%
- Mean
- 30.7%
- Median
- 9.1%
- Std dev
- 18.4%
- 10th percentile
- 36.2%
- 90th percentile
- 17.4%
- Min
This benchmark has a weight of 1.08 for chat models, measured over 7 models. The weight is how much it counts toward llminfo IQ. We measure it from the data, not from the raw scores.
Leaderboard
| # | Model | Score |
|---|---|---|
| 1 | Claude Sonnet 3.7GenericAgent· BrowserGym Leaderboard | 44.6% |
| 2 | Claude 3.5 SonnetGenericAgent· BrowserGym Leaderboard | 36.2% |
| 3 | GPT 4oGenericAgent· BrowserGym Leaderboard | 31.4% |
| 4 | GPT-4.1 miniGenericAgent· BrowserGym Leaderboard | 30.7% |
| 5 | Llama 3.1 405BGenericAgent· BrowserGym Leaderboard | 24% |
| 6 | Llama 3.1 70BGenericAgent· BrowserGym Leaderboard | 18.4% |
| 7 | GPT-4o miniGenericAgent· BrowserGym Leaderboard | 17.4% |
Data credits: BrowserGym Leaderboard (ServiceNow), MIT (licence)