AI BENCHY Category Failures
Instructions following: Wrong answer
Instructions following
Wrong answer
See which AI models are most likely to hit Wrong answer on Instructions following, so you can spot weak points faster. Sort by: Tests Correct ↓.
Failure Reasons
| Rank | Model | Company | Wrong answer Count | Category Score | Tests Correct | Response Time (avg) |
|---|---|---|---|---|---|---|
| #115 | Qwen3.5-27B none | Qwen | 1 | 6.3 | 1/2 | 1.03s |
| #116 | Hunter Alpha none | OpenRouter | 1 | 6.4 | 1/2 | 2.82s |
| #117 | Qwen3.5-35B-A3B none | Qwen | 1 | 6.3 | 1/2 | 809ms |
| #118 | Qwen3.6 27B none | Qwen | 1 | 6.2 | 1/2 | 1.92s |
| #120 | Mimo V2 PRO none | Xiaomi | 1 | 6.5 | 1/2 | 2.51s |
| #121 | Owl Alpha none | Openrouter | 1 | 6.4 | 1/2 | 2.63s |
| #122 | GLM 4.7 Flash none | Z.ai | 1 | 6.5 | 1/2 | 888ms |
| #123 | MiMo-V2.5-Pro none | Xiaomi | 1 | 6.4 | 1/2 | 1.03s |
| #124 | Kimi K2.6 none | Moonshot AI | 1 | 6.5 | 1/2 | 1.64s |
| #125 | GPT-5.4 none | OpenAI | 1 | 6.5 | 1/2 | 1.07s |
| #127 | Grok 4.20 none | X AI | 1 | 6.3 | 1/2 | 445ms |
| #128 | Qwen3.6 Flash none | Qwen | 1 | 6.3 | 1/2 | 1.10s |
| #131 | Qwen3.5-122B-A10B none | Qwen | 1 | 6.3 | 1/2 | 513ms |
| #132 | Mistral Small 4 medium | Mistral | 1 | 7.3 | 1/2 | 1.38s |
| #134 | GLM 5 Turbo none | Z.ai | 1 | 6.5 | 1/2 | 2.13s |