AGC-Bench - permpst: leaderboard
Metric: Dataset z-score. Source: huggingface.co. 83 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.4 | 2.14 |
| 2 | Claude Opus 4.6 | 1.94 |
| 3 | Claude Sonnet 4.6 | 1.51 |
| 4 | GLM-5 | 1.35 |
| 5 | Claude Opus 4.7 | 1.34 |
| 6 | GPT-5.5 | 1.27 |
| 7 | Qwen 3 14B | 1.27 |
| 8 | Claude Opus 4.5 | 1.18 |
| 9 | GPT-5.4 Mini | 1.06 |
| 10 | Kimi K2.6 | 0.96 |
| 11 | DeepSeek V3 (0324) | 0.91 |
| 12 | Claude Haiku 4.5 | 0.91 |
| 13 | Kimi K2.5 | 0.81 |
| 14 | pixtral-large-2411 | 0.81 |
| 15 | Qwen Turbo | 0.8 |
Interactive version: theaggregate.ai/benchmark?slug=agc-bench-permpst · How It Works · Data refreshed daily, snapshot 2026-09-05.