ALE-Bench: leaderboard
Score-based algorithmic programming benchmark built from AtCoder Heuristic Contest tasks, evaluating AI systems on hard optimization problems with hidden/private test evaluation.
Metric: Performance (Self-Refine x1) (imported). Source: sakanaai.github.io. Source: benchmarklist.com. Saturation forecast: Around June 2027. 89 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Best Expert | 2341 | |
| 2 | GPT-5.5 (xHigh) | 1942.97 | #31 (GPT-5.5) |
| 3 | GPT-5.3 Codex (xHigh) | 1655.22 | #69 (GPT-5.3 Codex) |
| 4 | GPT-5.4 (High) | 1607 | #79 (GPT-5.4) |
| 5 | GPT-5.4 (Medium) | 1520.72 | #79 (GPT-5.4) |
| 6 | Gemini 3 Flash (Preview) (High) | 1367.2 | #82 (Gemini 3 Flash (Preview)) |
| 7 | Claude Sonnet 4.6 (Medium) | 1327.3 | #90 (Claude Sonnet 4.6) |
| 8 | Claude Opus 4.7 (Non-reasoning) | 1323.05 | #49 (Claude Opus 4.7) |
| 9 | GPT-5.2 Codex (xHigh) | 1299.9 | #95 (GPT-5.2 Codex) |
| 10 | GPT-5.2 (High) | 1293.55 | #110 (GPT-5.2) |
| 11 | GPT-5.2 (Medium) | 1249.83 | #110 (GPT-5.2) |
| 12 | GPT-5.1 Codex (High) | 1244.92 | #159 (GPT-5.1 Codex) |
| 13 | Median Expert | 1231 | |
| 14 | GPT-5.1 (Thinking) | 1192.15 | #135 (GPT-5.1) |
| 15 | GPT-5.4 Mini (High) | 1188.58 | #212 (GPT-5.4 Mini) |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=ale-bench · How It Works · Data refreshed daily, snapshot 2026-10-20.