Benchmarks / Coding
LiveBench Coding
Contamination-limited monthly-refreshed benchmark; category mean over its tasks.
maintained by LiveBench · livebench.ai · 40 observations on record
Versions
1 version
| Version | Tasks |
|---|---|
| Release 2026-06-25 current livebench-coding-2026-06-25 | — |
Current leaderboard
Release 2026-06-25 · Accuracy · top 25 of 40 systems
| # | System | Accuracy % |
|---|---|---|
| 1 | Claude Fable 5max effort Anthropic | 86.0% |
| 2 | GPT 5.6 Solmax effort OpenAI | 83.9% |
| 3 | GPT 5.2 Codex OpenAI | 83.6% |
| 4 | GPT 5.6 Lunamax effort OpenAI | 82.9% |
| 5 | Claude Fable 5xhigh effort Anthropic | 82.5% |
| 6 | GPT 5.5 (2026-04-22)xhigh effort OpenAI | 82.2% |
| 7 | Claude Opus 4.7xhigh effort Anthropic | 82.1% |
| 8 | GPT 5.6 Solxhigh effort OpenAI | 81.8% |
| 9 | Claude Opus 4.8max effort Anthropic | 81.8% |
| 10 | Claude Sonnet 5xhigh effort Anthropic | 80.7% |
| 11 | Muse Spark 1.1high effort Meta AI | 80.4% |
| 12 | GPT 5.5 (2026-04-22)high effort OpenAI | 80.0% |
| 13 | GLM 5.2 Z.ai | 79.7% |
| 14 | Claude Opus 4.5 20251101high effort · 64K budget Anthropic | 79.7% |
| 15 | Claude Opus 4.8xhigh effort Anthropic | 79.3% |
| 16 | Claude Sonnet 4.6medium effort Anthropic | 79.3% |
| 17 | GPT 5.5 (2026-04-22) OpenAI | 78.6% |
| 18 | Kimi K2 6thinking Moonshot AI | 78.6% |
| 19 | GPT 5.6 Terramax effort OpenAI | 78.3% |
| 20 | Claude Opus 4.6high effort Anthropic | 78.2% |
| 21 | Gemini 3.5 Flashhigh effort Google DeepMind | 78.2% |
| 22 | Qwen3 6 Plus Alibaba | 78.2% |
| 23 | GPT 5.4 (2026-03-05)xhigh effort OpenAI | 77.5% |
| 24 | Muse Spark 1.1xhigh effort Meta AI | 77.2% |
| 25 | GPT 5.6 Lunaxhigh effort OpenAI | 76.7% |
One row per evaluated system — reasoning-effort variants rank separately.