CUDABeaver: leaderboard
Debugging CUDA programs has long been challenging because failures often arise from subtle interactions among hardware behavior, compiler decisions, memory hierarchy, and asynchronous execution. More importantly, with the rapid expansion of GPU usage across scientific computing, machine learning, graphics, and systems workloads, CUDA debugging has become more challenging than ever. Current evaluat
Metric: pass@5 (self-reported). Source: benchmarklist.com. Saturation forecast: Around July 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.4 | 29.1 | #79 |
| 2 | Qwen 3.6 Plus | 23 | #120 |
| 3 | Gemma 4 31B (IT) | 21.6 | #176 |
| 4 | Qwen 3.6 27B | 19.2 | #170 |
| 5 | Kimi K2.6 | 18.8 | #104 |
| 6 | GLM-4.7 | 12.7 | #192 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=cudabeaver · How It Works · Data refreshed daily, snapshot 2026-10-20.