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 301–379 of 379 benchmarks.
- Online-Mind2Web Medium (auto)3 models
- Personal Superintelligence Index3 models
- Search and Function-Calling3 models
- SWE-Atlas3 models
- Agent Startup Bench2 models
- BrowseComp Long Context 256k2 models
- Doubao Multi-Turn Bench2 models
- GDPval v22 models
- Gray Swan IPI Benchmark2 models
- Harvey LAB (Vals)2 models
- MobileMiniWob++_SR2 models
- Repo Env2 models
- SWE-bench Verified (Agentless)2 models
- Trae Error Fix2 models
- Web Bench2 models
- Agentic IF Index (Internal)1 model
- B3 AI Security Benchmark1 model
- BrowseComp-VL1 model
- CI Memories Coverage1 model
- CI Memories Violation Rate1 model
- Coding Agent Index1 model
- GDPval (Elo)1 model
- ImageMining1 model
- Mind2Web1 model
- MMSearch1 model
- OSWorld Extended1 model
- OSWorld Screenshot-only1 model
- OSWorld-G1 model
- SIFO1 model
- SIFO-Multiturn1 model
- Siren AgentDojo Utility1 model
- Tau3 Airline1 model
- Tau3 Retail1 model
- Tau3 Telecom1 model
- WebVoyager1 model