PerfCodeBench - Efficiency Gap Closed: leaderboard
Metric: Share of comparable tasks (%) whose correct code closes at least 80 percent of the baseline-to-reference performance gap (CGRE at least 0.8), 1,854 system-level performance-optimization tasks across C, C++, Go, Java, Python and CUDA; the model returns one replacement source file from the task metadata, interface contract and baseline in one shot; a candidate earns performance credit only if it compiles, runs and passes the task oracle; higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 20 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 72.91 |
| 2 | GPT-5.4 | 64.26 |
| 3 | Claude Opus 4.5 | 63.17 |
| 4 | Qwen 3.6 Max | 63.1 |
| 5 | DeepSeek V4 Pro | 62.88 |
| 6 | Kimi K2.6 | 61.7 |
| 7 | Qwen 3.6 27B | 61.39 |
| 8 | Gemini 3.1 Pro (Preview) | 60.67 |
| 9 | DeepSeek V4 Flash | 60.29 |
| 10 | Claude Sonnet 4.5 | 59.15 |
| 11 | Seed 2.0 Lite | 58.99 |
| 12 | Qwen 3.6 Plus | 54.85 |
| 13 | Kimi K2 | 43.58 |
| 14 | Qwen 3.6 35B A3B | 41.26 |
| 15 | Llama 4 Maverick | 40.26 |
Interactive version: theaggregate.ai/benchmark?slug=perfcodebench-efficiency-gap-closed · How It Works · Data refreshed daily, snapshot 2026-10-07.