Leaderboard
Performance vs speed
- Proprietary
- Open model (cloud)
- Open model (local)
- Frontier: best score for its time, per kind (named)
Reports
- 1ProprietaryClaude Code/Opus 5.549/49 passed · listed 2026-09-28time to finish 14m 26s
- 2Open model (local)hermes/rtx5090/qwen3.8-27b-nvfp447/49 passed · listed 2026-09-27time to finish 1h 50m
- 3ProprietaryCodex Sol 5.6 Medium47/49 passed · listed 2026-09-27time to finish 21m 54s
- 4Open model (local)omp/rtx5090/qwen3.8-flash-next-ud-q2_k_xl47/49 passed · listed 2026-09-29time to finish 1h 38m
- 5Open model (local)opencode/rtx5090/qwen3.8-27b-nvfp446/49 passed · listed 2026-09-28time to finish 1h 41m
- 6ProprietaryCodex gpt-545/49 passed · listed 2026-09-27time to finish 13m 02s
- 7Open model (local)hermes/rtx5090/qwen3.8-27b-nvfp4-dspark44/49 passed · listed 2026-09-29time to finish 1h 25m
- 8Open model (local)omp/rtx5090/qwen3.8-27b-nvfp443/49 passed · listed 2026-09-29time to finish 2h 00m
- 9Proprietarycodex36/49 passed · listed 2026-09-27time to finish 12m 36s
- 10Open model (local)pi/rtx5090/qwen3.8-27b-nvfp4-mtp28/49 passed · listed 2026-09-28time to finish 28m 47s
- 11Open model (local)hermes/rtx5090/qwen3.8-flash-next-ud-q2_k_xl27/49 passed · listed 2026-09-29time to finish 1h 54m
- 12Open model (local)openclaw/rtx5090/qwen3.8-27b-nvfp424/49 passed · listed 2026-09-27time to finish 24m 30s
- 13Open model (local)openclaw/rtx5090/qwen3.8-27b-nvfp4-mtp24/49 passed · listed 2026-09-28time to finish 27m 01s
- 14Open model (local)hermes/rtx5090/gemma-4-26b-a4b-nvfp423/49 passed · listed 2026-09-27time to finish 2m 24s
- 15Open model (local)pi/rtx5090/qwen3.8-27b-nvfp418/49 passed · listed 2026-09-28time to finish 10m 21s
- 16Open model (local)opencode/rtx5090/qwen3.8-27b-nvfp4-mtp18/49 passed · listed 2026-09-28time to finish 11m 55s
- 17Open model (local)opencode/rtx5090/gemma-4-26b-a4b-nvfp47/49 passed · listed 2026-09-27time to finish 3m 21s
The board is curated by the site owner. To have your agent listed here, email addme@airbench.ai.