PatternEval - Response-Pattern Failure Rate (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 share of final responses showing chain-of-thought leakage, repetition, contradiction or performative reasoning, as labelled by a Seed-2.0-Pro judge, in percent; lower is better.
Metric: Bad-pattern trigger rate (%, lower is better; share of final responses showing at least one of chain-of-thought leakage, repetition, logical contradiction or performative reasoning, labelled by a Seed-2.0-Pro judge with image access; 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: Estimated already saturated. 25 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | GPT-5.5 (Non-reasoning) | 5.04 | #49 (GPT-5.5) |
| 2 | GPT-5.4 (Non-reasoning) | 5.34 | #123 (GPT-5.4) |
| 3 | Claude Opus 4.8 | 22.06 | #56 |
| 4 | MiMo-V2.5 (Non-reasoning) | 24.92 | #354 (MiMo-V2.5) |
| 5 | Qwen 3.7 Max (Non-reasoning) | 26.77 | #145 (Qwen 3.7 Max) |
| 6 | Qwen 3.7 Plus (Non-reasoning) | 27.82 | #207 (Qwen 3.7 Plus) |
| 7 | Qwen 3.5 122B A10B (Non-reasoning) | 28.45 | #331 (Qwen 3.5 122B A10B) |
| 8 | Qwen 3.5 27B (Non-reasoning) | 29.32 | #359 (Qwen 3.5 27B) |
| 9 | Qwen 3.5 397B A17B (Non-reasoning) | 29.32 | #246 (Qwen 3.5 397B A17B) |
| 10 | Qwen 3.6 35B A3B (Non-reasoning) | 32.8 | #462 (Qwen 3.6 35B A3B) |
| 11 | Qwen 3.6 27B (Non-reasoning) | 32.81 | #341 (Qwen 3.6 27B) |
| 12 | Qwen 3.5 35B A3B (Non-reasoning) | 33 | #482 (Qwen 3.5 35B A3B) |
| 13 | Claude Opus 4.6 | 33.18 | #102 |
| 14 | Qwen 3 VL 32B Instruct | 35.03 | #585 |
| 15 | GPT-5 Nano (Minimal) | 35.67 | #889 (GPT-5 Nano) |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=patterneval-response-pattern-failure-rate-non-thinking · How It Works · Data refreshed daily, snapshot 2026-11-13.