PatternEval (Non-Thinking): leaderboard
PatternEval (2026) collects 2,415 multimodal prompts on which model responses often fail in user-visible ways, and runs hybrid-thinking models once with thinking on and once with it off. This board is non-thinking inference; the score is the accuracy of the final answer under each source task's criteria as judged by Qwen3-Max, in percent; higher is better.
Metric: Accuracy (%; correctness of the final user-visible answer under the source task's criteria, judged by Qwen3-Max; non-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 February 2027. 25 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Qwen 3.7 Max (Non-reasoning) | 67.07 | #145 (Qwen 3.7 Max) |
| 2 | Kimi K2.6 (Non-reasoning) | 66.47 | #178 (Kimi K2.6) |
| 3 | Qwen 3.7 Plus (Non-reasoning) | 66.28 | #207 (Qwen 3.7 Plus) |
| 4 | Qwen 3.5 397B A17B (Non-reasoning) | 61.96 | #246 (Qwen 3.5 397B A17B) |
| 5 | Kimi K2.5 (Non-reasoning) | 61.33 | #256 (Kimi K2.5) |
| 6 | Claude Opus 4.8 | 61.12 | #56 |
| 7 | GPT-5.4 (Non-reasoning) | 60.66 | #123 (GPT-5.4) |
| 8 | Claude Opus 4.6 | 59.45 | #102 |
| 9 | GPT-5.5 (Non-reasoning) | 59.2 | #49 (GPT-5.5) |
| 10 | Qwen 3.5 122B A10B (Non-reasoning) | 59.2 | #331 (Qwen 3.5 122B A10B) |
| 11 | Qwen 3.5 27B (Non-reasoning) | 59.07 | #359 (Qwen 3.5 27B) |
| 12 | Qwen 3.6 27B (Non-reasoning) | 57.33 | #341 (Qwen 3.6 27B) |
| 13 | Qwen 3.5 35B A3B (Non-reasoning) | 57.02 | #482 (Qwen 3.5 35B A3B) |
| 14 | Qwen 3.6 35B A3B (Non-reasoning) | 56.22 | #462 (Qwen 3.6 35B A3B) |
| 15 | MiMo-V2.5 (Non-reasoning) | 55.03 | #354 (MiMo-V2.5) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=patterneval-non-thinking · How It Works · Data refreshed daily, snapshot 2026-11-13.