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: Response Time (avg) ↑.
Failure Reasons
111/111
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) |
|---|---|---|---|---|---|---|---|
| #358 | Mercury 2.5 Preview default | Inception | 1 | 0.6 | $0.023 | 1/2 | 1.26s |
| #426 | Drex v1.5 default | Nace AI | 1 | 0.2 | $0.002 | 0/1 | 1.33s |
| #334 | Mistral Small 4 medium | Mistral | 1 | 0.7 | $0.126 | 1/2 | 1.38s |
| #262 | DeepSeek V4 Flash 0731 none | DeepSeek | 1 | 0.5 | $0.053 | 0/2 | 1.42s |
| #299 | Space Bunny Alpha low | Stealth | 1 | 0.4 | $0.000 | 0/2 | 1.43s |
| #233 | Gemini 3 Flash Preview none | 1 | 0.6 | $0.155 | 1/2 | 1.58s | |
| #341 | Mercury 2.5 Preview low | Inception | 1 | 0.6 | $0.015 | 1/2 | 1.60s |
| #307 | Kimi K2.6 none | Moonshot AI | 1 | 0.7 | $0.486 | 1/2 | 1.64s |
| #437 | Solar Decide Flash default | Upstage | 1 | 0.2 | $0.003 | 0/1 | 1.66s |
| #301 | Solar Pro 4 none | Upstage | 1 | 0.6 | $0.022 | 1/2 | 1.67s |
| #269 | Inkling none | Thinkingmachines | 1 | 0.6 | $0.292 | 1/2 | 1.72s |
| #327 | Qwen3.6 35B A3B none | Qwen | 1 | 0.6 | $0.105 | 1/2 | 1.86s |
| #246 | Qwen3.6 27B none | Qwen | 1 | 0.6 | $0.148 | 1/2 | 1.92s |
| #275 | Claude Sonnet 4.6 none | Anthropic | 1 | 0.7 | $0.661 | 1/2 | 1.96s |
| #367 | GLM 5V Turbo none | Z.ai | 1 | 0.7 | $0.052 | 1/2 | 1.97s |