PatternEval (Thinking): leaderboard
Metric: Accuracy (%; correctness of the final user-visible answer under the source task's criteria, judged by Qwen3-Max; thinking inference; PatternEval: 2,415 failure-enriched image-prompt pairs in three task families (visual perception and grounding, OCR and structured images, multimodal knowledge reasoning); official default decoding with only the reasoning control changed). Source: arxiv.org. Saturation forecast: Around January 2027. 25 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 | 74.9 |
| 2 | Kimi K2.6 (Thinking) | 73.19 |
| 3 | Seed 2.0 Pro (Thinking) | 72.98 |
| 4 | Qwen 3.7 Max (Thinking) | 72.3 |
| 5 | GPT-5.4 (Thinking) | 69.54 |
| 6 | Qwen 3.5 397B A17B (Thinking) | 68.93 |
| 7 | Claude Opus 4.8 (Thinking) | 67.7 |
| 8 | Claude Opus 4.6 (Thinking) | 67.63 |
| 9 | Kimi K2.5 (Thinking) | 67.06 |
| 10 | Qwen 3.5 122B A10B (Thinking) | 66.3 |
| 11 | Qwen 3.5 35B A3B (Thinking) | 65.07 |
| 12 | Qwen 3.5 27B (Thinking) | 64.05 |
| 13 | Qwen 3.6 35B A3B (Thinking) | 63.9 |
| 14 | Qwen 3.6 27B (Thinking) | 63.72 |
| 15 | Qwen 3.5 9B (Thinking) | 59.2 |
Interactive version: theaggregate.ai/benchmark?slug=patterneval-thinking · How It Works · Data refreshed daily, snapshot 2026-09-29.