Benchmarks
Agentic benchmarks
These benchmarks measure how well models carry out tasks that take several steps.
We track 379 benchmarks in this topic. Main benchmarks are listed by how many models they cover, most first. Where a benchmark has separate results for a subset of its tasks, those results appear right under it. Click one for its leaderboard.
Showing 201–300 of 379 benchmarks.
- BrowserGym WorkArena-L36 models
- finsearchcomp-global6 models
- finsearchcomp-greater-china6 models
- OSWorld 2.1 (strict)6 models
- BrowserGym AssistantBench5 models
- BrowserGym WebLINX5 models
- Harvey's Legal Agent Benchmark5 models
- Online-Mind2Web (auto-o4-mini)5 models
- Online-Mind2Web Easy5 models
- Online-Mind2Web Medium5 models
- Safeagents Judge Bench5 models
- BFCL v24 models
- BrowseComp Long Context 128k4 models
- HealthAdminBench4 models
- Terminal-Bench 4.0-AA4 models
- AgentBench3 models
- Connectors3 models
- FocusBench3 models
- LatchBio SingleCellBench3 models
- LatchBio SpatialBench Verified3 models
- MemoryBench3 models
- MM-Mind2Web3 models
- Online-Mind2Web (auto)3 models
- Online-Mind2Web Easy (auto)3 models
- Online-Mind2Web Hard (auto)3 models