LiveBench Code Generation: leaderboard
LiveBench's code generation task, where the model reads a problem statement and writes a complete program. Since the April 2025 refresh the questions are written around realistic use of real world libraries rather than taken from competition sites. Answers are checked automatically against objective ground truth with no model judge, and the score is a percentage where higher is better.
Metric: Score. Source: livebench.ai. Saturation forecast: Around February 2027. 71 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 5.5 (Max) | 95.78 | #11 (Claude Sonnet 5.5) |
| 2 | Claude Sonnet 5.5 (xHigh) | 92.96 | #11 (Claude Sonnet 5.5) |
| 3 | Claude Fable 5 (Max) | 91.55 | #16 (Claude Fable 5) |
| 4 | Claude Opus 5.5 (Max) | 91.55 | #3 (Claude Opus 5.5) |
| 5 | Claude Opus 5.5 (xHigh) | 91.55 | #3 (Claude Opus 5.5) |
| 6 | Claude Fable 5.1 (Max) | 90.14 | #9 (Claude Fable 5.1) |
| 7 | Claude Opus 4.7 (xHigh) | 85.92 | #49 (Claude Opus 4.7) |
| 8 | GPT-5.6 Sol (Max) | 83.1 | #19 (GPT-5.6 Sol) |
| 9 | GPT-6 Sol (Max) | 83.1 | #14 (GPT-6 Sol) |
| 10 | Muse Wonder (Max) | 83.1 | #15 (Muse Wonder) |
| 11 | Claude Sonnet 5 (xHigh) | 83.1 | #57 (Claude Sonnet 5) |
| 12 | GPT-5.5 (xHigh) | 81.69 | #31 (GPT-5.5) |
| 13 | Gemini 3.7 Flash (High) | 81.69 | #23 (Gemini 3.7 Flash) |
| 14 | Muse Spark 1.3 (xHigh) | 81.69 | #24 (Muse Spark 1.3) |
| 15 | Qwen 3.6 Plus | 80.28 | #120 |
No result here: #8 Gemini 4 Argon, #10 Muse Wonder Full.
Interactive version: theaggregate.ai/benchmark?slug=livebench-code-generation · How It Works · Data refreshed daily, snapshot 2026-10-20.