AI BENCHY Category Failures
Anti-AI Tricks: Did not follow instructions
Anti-AI Tricks
Did not follow instructions
See which AI models are most likely to hit Did not follow instructions on Anti-AI Tricks, so you can spot weak points faster.
Failure Reasons
| Rank | Model | Company | Did not follow instructions Count | Category Score | Tests Correct | Response Time (avg) |
|---|---|---|---|---|---|---|
| #17 | Gemini 3.1 Flash Lite Preview medium | 1 | 9.1 | 3/4 | 2.33s | |
| #29 | Gemini 3.1 Flash Lite Preview none | 1 | 7.5 | 2/4 | 1.04s | |
| #31 | GLM 5V Turbo medium | Z.ai | 1 | 7.2 | 2/4 | 10.8s |
| #34 | Kimi K2.6 medium | Moonshot AI | 1 | 7.0 | 2/4 | 11.6s |
| #36 | GPT-5.3 Chat none | OpenAI | 1 | 6.7 | 2/4 | 3.86s |
| #40 | GPT-5.2 medium | OpenAI | 1 | 6.5 | 2/4 | 7.81s |
| #45 | GPT-5 Mini medium | OpenAI | 1 | 7.1 | 2/4 | 13.9s |
| #54 | Mercury 2 medium | Inception | 1 | 6.9 | 2/4 | 1.12s |
| #68 | gpt-oss-120b medium | OpenAI | 1 | 6.7 | 2/4 | 10.2s |
| #71 | MiniMax M2.5 medium | Minimax | 1 | 7.9 | 2/4 | 20.8s |
| #80 | MiniMax M2.7 medium | Minimax | 1 | 7.9 | 2/4 | 40.3s |
| #84 | gpt-oss-120b none | OpenAI | 1 | 6.6 | 2/4 | 6.03s |
| #85 | Elephant none | Openrouter | 1 | 6.6 | 2/4 | 963ms |
| #87 | Qwen3 Coder Next none | Qwen | 1 | 3.6 | 0/4 | 3.31s |
| #92 | Qwen3 Coder Next medium | Qwen | 1 | 3.5 | 0/4 | 8.64s |