LiveBench Coding Average
Code · 2024-06-12
LiveBench leaderboard aggregate for livebench coding average. LiveBench is refreshed over time to reduce contamination and covers reasoning, coding, mathematics, language, data analysis, instruction following, and agentic coding.
Top models (higher is better)
| Model | Score |
|---|---|
| GPT-5.2-Codex | 83.6 |
| GPT-5.5 | 82.2 |
| Opus 4.7 | 82.1 |
| Sonnet 5 | 80.7 |
| Opus 4.5 | 79.7 |
| GLM-5.2 | 79.7 |
| Opus 4.8 | 79.3 |
| Sonnet 4.6 | 79.3 |
| Kimi K2.6 | 78.6 |
| Opus 4.6 | 78.2 |
| Gemini 3.5 Flash | 78.2 |
| Qwen3.6 Plus (2026-04-02) | 78.2 |
| GPT-5.4 | 77.5 |
| Gemini 3.1 Pro Preview | 76.5 |
| GPT-5.2 | 76.1 |
| Qwen3.7-Max (source unspecified) | 74.2 |
| Kimi K2.7 Code | 74.0 |
| Qwen3.6 27B | 71.8 |
| GPT-5.4 Mini | 71.6 |
| GPT-5.4 Nano | 70.8 |
| DeepSeek-V4-Pro | 70.0 |
| Grok 4.3 | 69.9 |
| DeepSeek-V4-Flash | 69.2 |
| MiniMax M3 | 68.2 |
| Grok Build 0.1 | 65.4 |