CCBench: leaderboard
About 180 feature tasks in private CodeCrafters codebases (e.g. add authentication to a Redis clone, backreferences to a regex engine), graded by CodeCrafters' test runners; success rate (2026).
Metric: Accuracy (%). Source: ccbench.org. Source: huggingface.co. Saturation forecast: Estimated already saturated. 229 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | InternVL3-78B | 83.5 | #352 |
| 2 | InternVL3-38B | 80.6 | #405 |
| 3 | InternVL3-8B | 77.8 | #620 |
| 4 | InternVL2-8B | 77.1 | #817 |
| 5 | InternVL3-14B | 77.1 | #505 |
| 6 | GPT-4.1 (2025-04-14) | 73.5 | #250 |
| 7 | Qwen 2 VL 72B | 69.8 | #460 |
| 8 | Qwen 2 VL 7B | 65.7 | #750 |
| 9 | GPT-4.1 Mini | 64.3 | #355 |
| 10 | Moonshot v1 8K | 60.2 | #617 |
| 11 | Qwen 2.5 VL 7B | 59.2 | #643 |
| 12 | Grok 2 (1212) | 57.3 | #496 |
| 13 | Claude 3.5 Sonnet (20241022) | 56.1 | #291 |
| 14 | MiniCPM-V-2.6 | 55.7 | #836 |
| 15 | Claude 3.7 Sonnet | 55.1 | #249 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=ccbench · How It Works · Data refreshed daily, snapshot 2026-10-20.