Benchmark
BrowserGym MiniWoB
Agentic benchmarks measure how well models carry out tasks that take several steps. Scores are percentages.
All agentic benchmarksScores from BrowserGym Leaderboard.
- Models measured
- 11
- Scores recorded
- 11
- Sources
- 1
- IQ weight
- 0.99
Score distribution
- 65.5%
- Mean
- 64.8%
- Median
- 4.9%
- Std dev
- 57.6%
- 10th percentile
- 71%
- 90th percentile
- 56.6%
- Min
This benchmark has a weight of 0.99 for chat models, measured over 11 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 | GPT 5GenericAgent· BrowserGym Leaderboard | 71.5% |
| 2 | GPT-5 miniGenericAgent· BrowserGym Leaderboard | 71% |
| 3 | Claude Sonnet 4GenericAgent· BrowserGym Leaderboard | 70.7% |
| 4 | Claude 3.5 SonnetGenericAgent· BrowserGym Leaderboard | 69.8% |
| 5 | GPT OSS 120BGenericAgent· BrowserGym Leaderboard | 66.4% |
| 6 | GPT-5 nanoGenericAgent· BrowserGym Leaderboard | 64.8% |
| 7 | Llama 3.1 405BGenericAgent· BrowserGym Leaderboard | 64.6% |
| 8 | GPT OSS 20bGenericAgent· BrowserGym Leaderboard | 64% |
| 9 | GPT 4oGenericAgent· BrowserGym Leaderboard | 63.8% |
| 10 | Llama 3.1 70BGenericAgent· BrowserGym Leaderboard | 57.6% |
| 11 | GPT-4o miniGenericAgent· BrowserGym Leaderboard | 56.6% |
Data credits: BrowserGym Leaderboard (ServiceNow), MIT (licence)