Benchmarks · reasoning
FrontierML
Uses the published mean normalized task score across wallet-budget settings; the index selects each model's highest-scoring setting. Score: mean normalized task score. Collected by Jake Boggs' benchmark dashboard.
18 models · 18 results · data as of 2026-11-06 · frontierml.org ↗
Leaderboard
Opus 5.5 max
All results
18 rows| # | Model | Effort | FrontierML |
|---|---|---|---|
| 1 | Opus 5.5 | max | 58.8% |
| 2 | Sonnet 5.5 | max | 50.4% |
| 3 | GPT-6 Astra | max | 44.5% |
| 4 | Fable 5.1 | max | 41.9% |
| 5 | Gemini 4 Argon | max | 39.5% |
| 6 | GPT-6.1 Sol | max | 38.7% |
| 7 | Haiku 5.5 | max | 34.9% |
| 8 | Kimi K3.1 | max | 29.6% |
| 9 | MiniMax M3.1 | max | 24.0% |
| 10 | DeepSeek V4.1 Flash | max | 23.5% |
| 11 | Muse Wonder (Max) | max | 21.6% |
| 12 | Kimi K3 | max | 20.3% |
| 13 | Qwen 3.8 Max | max | 19.9% |
| 14 | GLM 5.3 | max | 19.6% |
| 15 | Muse Spark 1.3 | max | 16.0% |
| 16 | GPT-6 Luna | max | 13.3% |
| 17 | Grok 4.7 | max | 12.2% |
| 18 | Gemini 3.8 Flash | max | 9.5% |