AI BENCHY Category Failures
Puzzle Solving: Wrong answer
Puzzle Solving
Wrong answer
See which AI models are most likely to hit Wrong answer on Puzzle Solving, so you can spot weak points faster.
Failure Reasons
| Rank | Model | Company | Wrong answer Count | Category Score | Tests Correct | Response Time (avg) |
|---|---|---|---|---|---|---|
| #76 | Kimi K2.5 none | Moonshot AI | 3 | 3.1 | 0/3 | 4.73s |
| #87 | Qwen3 Coder Next none | Qwen | 3 | 3.2 | 0/3 | 22.9s |
| #89 | GPT-4o-mini none | OpenAI | 3 | 3.7 | 0/3 | 1.30s |
| #91 | Mercury 2 none | Inception | 3 | 3.1 | 0/3 | 533ms |
| #94 | MiMo-V2-Flash none | Xiaomi | 3 | 3.6 | 0/3 | 1.38s |
| #95 | Grok 4.1 Fast none | X AI | 3 | 3.2 | 0/3 | 1.28s |
| #59 | Qwen3.5-Flash none | Qwen | 2 | 3.3 | 0/3 | 5.90s |
| #61 | Seed-2.0-Lite none | Bytedance Seed | 2 | 5.2 | 1/3 | 2.46s |
| #63 | Qwen3.5-35B-A3B none | Qwen | 2 | 3.9 | 0/3 | 1.34s |
| #70 | Qwen3.5-122B-A10B none | Qwen | 2 | 5.4 | 1/3 | 982ms |
| #78 | Trinity Large Preview none | Arcee AI | 2 | 5.4 | 1/3 | 3.30s |
| #85 | Elephant none | Openrouter | 2 | 3.3 | 0/3 | 849ms |
| #93 | GLM 4.7 Flash medium | Z.ai | 2 | 2.9 | 0/3 | 12.9s |
| #96 | GPT-5.4 Nano none | OpenAI | 2 | 3.7 | 0/3 | 1.29s |
| #11 | Gemini 3.1 Flash Lite Preview high | 1 | 7.7 | 2/3 | 46.3s |