AI BENCHY Category
Instructions following Ranking
See which AI models perform best on Instructions following, which ones stay reliable, and where the biggest gaps appear. Sort by: Tests Correct ↓.
| Rank | Model | Company | Instructions following Score | Score | Tests Correct | Response Time (avg) |
|---|---|---|---|---|---|---|
| #134 | GLM 5 Turbo none | Z.ai | 6.5 | 5.2 | 1/2 | 2.13s |
| #135 | Kimi K2.5 none | Moonshot AI | 6.5 | 5.2 | 1/2 | 2.67s |
| #139 | DeepSeek V4 Flash none | DeepSeek | 6.5 | 5.0 | 1/2 | 17.5s |
| #140 | Qwen3 Coder Next none | Qwen | 6.3 | 4.9 | 1/2 | 7.78s |
| #141 | Nemotron 3 Super none | NVIDIA | 6.3 | 4.9 | 1/2 | 804ms |
| #142 | Mistral Small 4 none | Mistral | 6.5 | 4.9 | 1/2 | 380ms |
| #143 | MiMo-V2.5 none | Xiaomi | 6.5 | 4.9 | 1/2 | 751ms |
| #144 | GPT-5.4 Mini none | OpenAI | 6.3 | 4.9 | 1/2 | 728ms |
| #145 | Laguna M.1 none | Poolside | 6.3 | 4.8 | 1/2 | 683ms |
| #146 | Laguna Xs.2 none | Poolside | 6.5 | 4.8 | 1/2 | 439ms |
| #147 | GPT-4o-mini none | OpenAI | 6.3 | 4.8 | 1/2 | 1.11s |
| #148 | GPT-5.4 Nano none | OpenAI | 6.3 | 4.7 | 1/2 | 784ms |
| #149 | Nemotron 3 Nano Omni 30b A3b Reasoning medium | NVIDIA | 7.3 | 4.6 | 1/2 | 1.37s |
| #150 | Qwen3 Coder Next medium | Qwen | 6.3 | 4.6 | 1/2 | 7.49s |
| #152 | MiMo-V2-Flash none | Xiaomi | 6.5 | 4.6 | 1/2 | 857ms |