Benchmarks · coding
ALE-Bench
Uses each model's best all-scenario performance mean, normalized to the current leaderboard maximum. Score: best all-scenario performance mean. Collected by Jake Boggs' benchmark dashboard.
122 models · 131 results · data as of 2026-11-06 · sakanaai.github.io ↗
Leaderboard
GPT-6 Astra max
All results
131 rows| # | Model | Effort | ALE-Bench |
|---|---|---|---|
| 1 | GPT-6 Astra | max | 3164.4 |
| 2 | GPT-6.1 Sol | max | 2989.8 |
| 3 | GPT-6.1 Terra | max | 2875.0 |
| 4 | Fable 5.1 | max | 2849.2 |
| 5 | Gemini 4 Argon | high | 2804.5 |
| 6 | Opus 5.5 | max | 2748.1 |
| 7 | GPT-6 Sol | max | 2636.6 |
| 8 | Opus 5 | max | 2622.5 |
| 9 | Gemini 4 Argon | low | 2599.2 |
| 10 | Fable 5 | max | 2463.2 |
| 11 | Sonnet 5.5 | max | 2448.3 |
| 12 | GPT-5.6 Sol | max | 2409.5 |
| 13 | Gemini 3.8 Flash | high | 2181.6 |
| 14 | GPT-5.6 Terra | max | 2154.1 |
| 15 | Kimi K3.1 | max | 2131.7 |
| 16 | GPT-5.5 | xhigh | 2108.3 |
| 17 | Opus 4.8 | max | 2101.5 |
| 18 | Gemini 3.1 Pro | high | 2041.2 |
| 19 | DeepSeek V4.1 Flash 1017 | max | 2011.8 |
| 20 | Haiku 5.5 | max | 1995.7 |
| 21 | Kimi K3 | max | 1990.6 |
| 22 | Muse Spark 1.3 | max | 1980.8 |
| 23 | Qwen 3.8 Max | max | 1964.0 |
| 24 | GPT-5.5 | medium | 1938.3 |
| 25 | GLM 5.3 | high | 1936.5 |
| 26 | GPT-5.3 Codex | xhigh | 1934.3 |
| 27 | Opus 4.7 | max | 1897.1 |
| 28 | Gemini 3.7 Flash | high | 1876.3 |
| 29 | Muse Spark Open | xhigh | 1829.7 |
| 30 | DeepSeek V4.1 Flash | max | 1828.8 |