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. Sort by: Response Time (avg) ↑.
Failure Reasons
246/246
Filter models
No models match the current search and filters.
| Rank | Model | Company | Wrong answer Count | Category Score | Total Cost | Tests Correct | Response Time (avg) |
|---|---|---|---|---|---|---|---|
| #274 | Beam 1.0 501B A23B high | Reflection | 1 | 0.8 | $0.674 | 2/3 | 4.02s |
| #333 | Kimi K2.5 none | Moonshot AI | 3 | 0.3 | $0.262 | 0/3 | 4.04s |
| #151 | GPT-5.6 Luna medium | OpenAI | 1 | 0.8 | $0.065 | 2/3 | 4.04s |
| #184 | GPT-5.2 Chat default | OpenAI | 1 | 0.8 | $0.594 | 2/3 | 4.10s |
| #381 | Owl Alpha default | Openrouter | 1 | 0.5 | $0.000 | 1/3 | 4.18s |
| #307 | Gemma 4 31B none | 1 | 0.6 | $0.031 | 1/3 | 4.23s | |
| #232 | Beam 1.0 501B A23B xhigh | Reflection | 2 | 0.6 | $0.688 | 1/3 | 4.46s |
| #419 | Hy3 preview none | Tencent | 2 | 0.3 | $0.007 | 0/3 | 4.56s |
| #261 | Claude Opus 4.6 medium | Anthropic | 1 | 0.8 | $2.961 | 2/3 | 4.71s |
| #128 | GLM 5.3 high | Z.ai | 1 | 0.8 | $0.495 | 2/3 | 4.74s |
| #145 | Qwen3.8 27B low | Qwen | 1 | 0.8 | ~$0.089 | 2/3 | 4.91s |
| #41 | Muse Wonder low | Meta | 1 | 0.6 | $2.637 | 1/3 | 5.08s |
| #253 | Qwen3.6 27B none | Qwen | 1 | 0.5 | $0.148 | 1/3 | 5.15s |
| #400 | Grok 4.20 Multi Agent Beta medium | X AI | 1 | 0.7 | $5.599 | 1/3 | 5.19s |
| #286 | Inkling Small medium | Thinkingmachines | 1 | 0.6 | $0.113 | 1/3 | 5.24s |