LiveBench Code Generation: leaderboard
Metric: Score. Source: livebench.ai. 54 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Claude Fable 5 (Max) | 91.55 |
| 2 | Claude Fable 5.1 (Max) | 90.14 |
| 3 | Claude Opus 4.7 (xHigh) | 85.92 |
| 4 | GPT-5.6 Sol (Max) | 83.1 |
| 5 | Claude Sonnet 5 (xHigh) | 83.1 |
| 6 | GPT-5.5 (xHigh) | 81.69 |
| 7 | Gemini 3.7 Flash (High) | 81.69 |
| 8 | Muse Spark 1.3 (xHigh) | 81.69 |
| 9 | Qwen 3.6 Plus | 80.28 |
| 10 | Kimi K3 | 80.28 |
| 11 | Gemini 3.5 Flash (High) | 80.28 |
| 12 | Claude Opus 5 (Max) | 80.28 |
| 13 | GPT-6 (Max) | 80.28 |
| 14 | GPT-5.2 Codex | 80.28 |
| 15 | Claude Opus 4.6 (Thinking, High) | 80.28 |
Interactive version: theaggregate.ai/benchmark?slug=livebench-code-generation · How It Works · Data refreshed daily, snapshot 2026-09-05.