TurkBench - Instruction Following: leaderboard
Metric: LLM-as-a-judge score (0-100). Source: huggingface.co. 38 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | DeepSeek V3.2 Exp | 96.8 |
| 2 | DeepSeek V3.1 | 94.92 |
| 3 | Qwen 3 Next 80B A3B Instruct | 94.27 |
| 4 | GLM-5 FP8 | 94.08 |
| 5 | GPT-OSS-120B | 93.6 |
| 6 | Qwen 3.5 397B A17B | 93.42 |
| 7 | GLM-4.6 | 92.58 |
| 8 | Qwen 3 30B A3B 2507 Instruct | 92.51 |
| 9 | Qwen 3.5 35B A3B | 92.39 |
| 10 | Qwen 3.5 27B | 91.82 |
| 11 | Qwen 3.5 122B A10B | 91.54 |
| 12 | Step 3.5 Flash | 91.07 |
| 13 | MiMo-V2-Flash | 90.98 |
| 14 | MiniMax-M2 | 90.88 |
| 15 | MiniMax-M2.5 | 89.29 |
Interactive version: theaggregate.ai/benchmark?slug=turkbench-instruction-following · How It Works · Data refreshed daily, snapshot 2026-09-19.