AI BENCHY Category Failures
Combined
Wrong answer
Combined
Wrong answer
See which AI models are most likely to hit Wrong answer on Combined, so you can spot weak points faster. Sort by: Tests Correct ↑.
Related Failure Reasons
| Rank | Model | Company | Wrong answer Count | Category Score | Tests Correct | Response Time (avg) |
|---|---|---|---|---|---|---|
| #5 | Gemini 3 Flash Preview low | 1 | 10.0 | 0/1 | 3.27s | |
| #6 | Gemini 3 Pro Preview medium | 1 | 10.0 | 0/1 | 10.4s | |
| #17 | Gemini 3.1 Flash Lite Preview low | 1 | 10.0 | 0/1 | 11.9s | |
| #20 | Gemini 3 Flash Preview none | 1 | 10.0 | 0/1 | 3.56s | |
| #22 | Gemini 3.1 Flash Lite Preview none | 1 | 10.0 | 0/1 | 3.20s | |
| #29 | Qwen3.5 Plus 2026-02-15 none | Qwen | 1 | 10.0 | 0/1 | 6.65s |
| #31 | GLM 5 none | Z.ai | 1 | 10.0 | 0/1 | 4.98s |
| #37 | Qwen3.5-Flash none | Qwen | 1 | 10.0 | 0/1 | 6.22s |
| #38 | Gemini 2.5 Flash none | 1 | 10.0 | 0/1 | 4.39s | |
| #40 | Qwen3.5-122B-A10B none | Qwen | 1 | 10.0 | 0/1 | 46.0s |
| #41 | Qwen3.5-27B none | Qwen | 1 | 10.0 | 0/1 | 9.39s |
| #42 | Qwen3.5-35B-A3B none | Qwen | 1 | 10.0 | 0/1 | 47.4s |
| #44 | GPT-5.4 none | OpenAI | 1 | 10.0 | 0/1 | 2.89s |
| #45 | Trinity Large Preview none | Arcee AI | 1 | 10.0 | 0/1 | 8.91s |
| #46 | Kimi K2.5 none | Moonshot AI | 1 | 10.0 | 0/1 | 19.2s |
| #47 | GPT-4o-mini none | OpenAI | 1 | 10.0 | 0/1 | 7.58s |
| #48 | Qwen3 Coder Next none | Qwen | 1 | 10.0 | 0/1 | 45.1s |
| #50 | Qwen3 Coder Next medium | Qwen | 1 | 10.0 | 0/1 | 4.28s |
| #51 | Mercury 2 none | Inception | 1 | 10.0 | 0/1 | 606ms |
| #53 | Grok 4.1 Fast none | X AI | 1 | 10.0 | 0/1 | 3.33s |
| #54 | MiMo-V2-Flash none | Xiaomi | 1 | 10.0 | 0/1 | 2.87s |