ALE-Bench
Code · 2025-06-17
ALE-Bench evaluates coding agents on improving algorithmic solutions under an execution-based optimization harness. The leaderboard reports aggregate performance scores, with higher values indicating better solutions.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-6 Astra | 3164 |
| GPT-6.1 Sol | 2990 |
| Claude Fable 5.1 | 2849 |
| Gemini 4 Argon | 2805 |
| Claude Opus 5.5 | 2748 |
| GPT-6 Sol | 2637 |
| Claude Opus 5 | 2622 |
| Claude Fable 5 | 2463 |
| Claude Sonnet 5.5 | 2448 |
| GPT-5.6 Sol | 2410 |
| Gemini 3.8 Flash | 2182 |
| GPT-5.6 Terra | 2154 |
| GPT-5.5 | 2108 |
| Opus 4.8 | 2102 |
| Gemini 3.1 Pro Preview | 2041 |
| DeepSeek V4.1 Flash 1017 | 2012 |
| Kimi K3 | 1991 |
| Muse Spark 1.3 | 1981 |
| GLM-5.3 | 1937 |
| GPT-5.3-Codex | 1934 |
| Opus 4.7 | 1897 |
| Gemini 3.7 Flash | 1876 |
| Muse Spark Open | 1830 |
| DeepSeek V4.1 Flash | 1829 |
| GPT-5.6 Luna | 1829 |
| Gemini 3.6 Flash | 1825 |
| Grok 4.6 | 1822 |
| GPT-5.4 | 1811 |
| GPT-6 Luna | 1791 |
| Sonnet 5 | 1783 |
| Sonnet 4.6 | 1768 |
| DeepSeek V4 Pro 0813 | 1730 |
| Gemini 3 Pro Preview | 1729 |
| Gemini 3 Flash Preview | 1722 |
| GPT-5.2 | 1715 |
| Gemini 3.5 Flash | 1712 |
| Kimi K2.6 | 1695 |
| Mercury 3 | 1692 |
| DeepSeek-V4-Flash-0731 | 1686 |
| GPT-5.2-Codex | 1686 |