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: Failure Count ↑.
Failure Reasons
| Rank | Model | Company | Wrong answer Count | Category Score | Tests Correct | Response Time (avg) |
|---|---|---|---|---|---|---|
| #21 | Gemini 3 Flash Preview none | 1 | 6.4 | 1/2 | 1.58s | |
| #28 | GPT-5.2 Chat none | OpenAI | 1 | 7.5 | 1/2 | 5.46s |
| #33 | GLM 5.1 medium | Z.ai | 1 | 6.4 | 1/2 | 7.47s |
| #36 | GPT-5.3 Chat none | OpenAI | 1 | 8.3 | 1/2 | 3.29s |
| #42 | Claude Sonnet 4.6 none | Anthropic | 1 | 6.5 | 1/2 | 1.96s |
| #48 | Gemma 4 31B none | 1 | 6.5 | 1/2 | 2.84s | |
| #55 | MiMo-V2-Omni none | Xiaomi | 1 | 6.5 | 1/2 | 4.18s |
| #58 | GLM 5V Turbo none | Z.ai | 1 | 6.5 | 1/2 | 1.97s |
| #59 | Qwen3.5-Flash none | Qwen | 1 | 6.3 | 1/2 | 8.81s |
| #60 | Gemma 4 26B A4B none | 1 | 4.4 | 0/2 | 1.08s | |
| #62 | Gemini 2.5 Flash none | 1 | 8.0 | 1/2 | 672ms | |
| #63 | Qwen3.5-35B-A3B none | Qwen | 1 | 6.3 | 1/2 | 809ms |
| #65 | MiMo-V2-Pro none | Xiaomi | 1 | 6.5 | 1/2 | 2.51s |
| #66 | GPT-5.4 none | OpenAI | 1 | 6.5 | 1/2 | 1.07s |
| #69 | Kimi K2.6 none | Moonshot AI | 1 | 6.5 | 1/2 | 1.64s |