Benchmarks · reasoning
FrontierCS 2
Uses the published pass rate against the human reference on every hidden metric, not the separate FCI index. Score: pass rate beating the human reference on every metric. Collected by Jake Boggs' benchmark dashboard.
17 models · 17 results · data as of 2026-11-06 · frontier-cs.org ↗
Leaderboard
Opus 5.5 max
All results
17 rows| # | Model | Effort | FrontierCS 2 |
|---|---|---|---|
| 1 | Opus 5.5 | max | 58.3% |
| 2 | Gemini 4 Carbon | max | 54.2% |
| 3 | Gemini 4 Argon | max | 54.2% |
| 4 | GPT-6 Astra | max | 45.8% |
| 5 | GPT-6.1 Terra | max | 39.6% |
| 6 | GPT-6.1 Sol | max | 37.5% |
| 7 | Fable 5.1 | max | 35.4% |
| 8 | Muse Wonder (Max) | max | 22.9% |
| 9 | GLM 5.3 | max | 22.9% |
| 10 | Kimi K3.1 | max | 20.8% |
| 11 | Kimi K3 | max | 20.8% |
| 12 | DeepSeek V4.1 Flash 1017 | max | 14.6% |
| 13 | DeepSeek V4.1 Flash | max | 12.5% |
| 14 | GLM 5.3 Flash | max | 12.5% |
| 15 | Muse Spark 1.3 | max | 12.5% |
| 16 | Qwen 3.8 Max | max | 10.4% |
| 17 | MiniMax M3.1 | max | 8.3% |