PassBench: leaderboard
Modern tensor compilers such as TorchInductor deliver substantial speedups on mainstream models, yet face a systematic performance ceiling on long-tail workloads -- our profiling shows that 43% of real-world subgraphs experience end-to-end slowdowns under default compilation. While LLMs offer a path toward automated optimization, existing efforts focus on standalone kernel generation. We argue tha
Metric: AS Score (self-reported). Source: benchmarklist.com. Saturation forecast: Around July 2027. 11 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Sonnet 4.6 | 44.8 | #90 |
| 2 | GPT-5.4 | 41 | #79 |
| 3 | Claude Opus 4.6 | 41 | #63 |
| 4 | GLM-5.1 | 24 | #97 |
| 5 | MiniMax-M2.7 | 20.8 | #258 |
| 6 | Qwen 3 30B A3B | 13.9 | #493 |
| 7 | Qwen 3 4B | 10.8 | #829 |
No result here: #3 Claude Opus 5.5, #6 GPT-6 Astra, #7 GPT-6.1 Sol.
Interactive version: theaggregate.ai/benchmark?slug=passbench · How It Works · Data refreshed daily, snapshot 2026-10-20.