TRACK - Code Generation (Closed Book): leaderboard
Metric: Holistic pass (%; BigCodeBench-derived coding problems that require specific library API versions, 500 items; the final answer is correct and the reasoning entails every required fact; answered closed book, with no facts injected). Source: arxiv.org. Saturation forecast: Estimated already saturated. 7 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-4.1 Mini | 35.9 |
| 2 | Qwen 3 8B | 27.2 |
| 3 | Qwen 3 4B | 24.7 |
| 4 | Qwen 3 1.7B | 17.4 |
Interactive version: theaggregate.ai/benchmark?slug=track-code-generation-closed-book · How It Works · Data refreshed daily, snapshot 2026-09-26.