Benchmarks / Coding
LiveCodeBench
Competitive-programming problems with rolling contamination-free windows.
maintained by LiveCodeBench · livecodebench.github.io · 190 observations on record
Versions
2 versions
| Version | Tasks |
|---|---|
| LiveCodeBench v6 current livecodebench-v6 | 1,055 |
| LiveCodeBench v5 livecodebench-v5 | — |
Current leaderboard
LiveCodeBench v6 · pass@1 · top 25 of 28 systems
| # | System | pass@1 % |
|---|---|---|
| 1 | O4 Mini (2025-04-16)high effort OpenAI | 80.2% |
| 2 | O3 (2025-04-16)high effort OpenAI | 75.8% |
| 3 | O4 Mini (2025-04-16)medium effort OpenAI | 74.2% |
| 4 | Gemini 2.5 Pro 06.05 Google DeepMind | 73.6% |
| 5 | Deepseek R1 0528 DeepSeek | 73.1% |
| 6 | Gemini 2.5 Pro 05.06 Google DeepMind | 71.8% |
| 7 | Exaone 4.0 32B Unknown | 70.0% |
| 8 | Openreasoning Nemotron 32B Unknown | 69.8% |
| 9 | O3 Mini (2025-01-31)high effort OpenAI | 67.4% |
| 10 | Opencodereasoning Nemotron 1.1 32B Unknown | 66.8% |
| 11 | Grok 3 Minihigh effort xAI | 66.7% |
| 12 | O4 Mini (2025-04-16)low effort OpenAI | 65.9% |
| 13 | Qwen3 235B A22b Alibaba | 65.9% |
| 14 | Xbai O4medium effort Unknown | 65.0% |
| 15 | O3 Mini (2025-01-31)medium effort OpenAI | 63.0% |
| 16 | Gemini 2.5 Flash 05.20 Google DeepMind | 61.9% |
| 17 | Gemini 2.5 Flash 04.17 Google DeepMind | 60.6% |
| 18 | O3 Mini (2025-01-31)low effort OpenAI | 57.0% |
| 19 | Claude Opus 4thinking Anthropic | 56.6% |
| 20 | Claude Sonnet 4thinking Anthropic | 56.0% |
| 21 | Claude Sonnet 4 Anthropic | 47.1% |
| 22 | Claude Opus 4 Anthropic | 46.9% |
| 23 | Claude 3.5 Sonnet 20241022 Anthropic | 36.4% |
| 24 | GPT 4o (2024-08-06) OpenAI | 29.5% |
| 25 | GPT 4 Turbo (2024-04-09) OpenAI | 28.7% |
One row per evaluated system — reasoning-effort variants rank separately.