BigCodeBench — leaderboard
Challenging coding benchmark with 1,140 function-level tasks requiring composition of multiple function calls and complex instructions.
Metric: Pass@1 (%). Source: bigcode-bench.github.io. Status: saturation imminent. 126 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Human Expert | 97 |
| 2 | GPT-4o (2024-05-13) | 51.1 |
| 3 | DeepSeek V3 | 50 |
| 4 | Llama 4 Maverick | 49.7 |
| 5 | Qwen 2.5 Coder 32B Instruct | 49 |
| 6 | GPT-4.1 Mini | 48.9 |
| 7 | GPT-4 Turbo | 48.2 |
| 8 | Qwen 2.5 Coder 14B Instruct | 48.2 |
| 9 | GPT-4o (2024-11-20) | 48 |
| 10 | Athene-V2-Chat | 47.2 |
| 11 | Llama 3.3 70B Instruct | 46.9 |
| 12 | Claude 3.5 Sonnet (20240620) | 46.8 |
| 13 | Llama 3.1 70B Instruct | 46.1 |
| 14 | GPT-4o Mini (2024-07-18) | 46.1 |
| 15 | Claude 3.5 Haiku (20241022) | 46.1 |
Interactive version: theaggregate.ai/benchmark?slug=bigcodebench · How the rankings work · Data refreshed daily, snapshot 2026-07-22.