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: Response Time (avg) ↓.
| Rank | Model | Company | Instructions following Score | Score | Tests Correct | Response Time (avg) |
|---|---|---|---|---|---|---|
| #42 | Claude Sonnet 4.6 none | Anthropic | 6.5 | 7.4 | 1/2 | 1.96s |
| #17 | Gemini 3.1 Flash Lite Preview medium | 10.0 | 8.2 | 2/2 | 1.91s | |
| #38 | GPT-5.4 Nano medium | OpenAI | 9.8 | 7.6 | 2/2 | 1.88s |
| #49 | Qwen3.5 Plus 2026-02-15 none | Qwen | 10.0 | 6.8 | 2/2 | 1.67s |
| #69 | Kimi K2.6 none | Moonshot AI | 6.5 | 5.8 | 1/2 | 1.64s |
| #75 | GLM 5.1 none | Z.ai | 8.3 | 5.6 | 1/2 | 1.58s |
| #21 | Gemini 3 Flash Preview none | 6.4 | 8.1 | 1/2 | 1.58s | |
| #3 | Claude Opus 4.7 medium | Anthropic | 10.0 | 9.2 | 2/2 | 1.57s |
| #64 | DeepSeek V3.2 none | DeepSeek | 10.0 | 6.1 | 2/2 | 1.52s |
| #88 | Nemotron 3 Super none | NVIDIA | 4.9 | 5.1 | 0/2 | 1.50s |
| #22 | Gemini 3.1 Flash Lite Preview low | 10.0 | 8.1 | 2/2 | 1.49s | |
| #53 | GLM 5 none | Z.ai | 10.0 | 6.6 | 2/2 | 1.48s |
| #4 | Claude Opus 4.7 none | Anthropic | 10.0 | 9.2 | 2/2 | 1.46s |
| #73 | Mistral Small 4 medium | Mistral | 7.3 | 5.7 | 1/2 | 1.38s |
| #89 | GPT-4o-mini none | OpenAI | 4.8 | 4.9 | 0/2 | 1.27s |