AGI-Eval Community - Instruction Following: leaderboard
Metric: Accuracy (%). Source: agi-eval.cn. 141 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5.5 | 71.29 |
| 2 | GPT-5.4 | 69.47 |
| 3 | GLM-5.2 | 69.23 |
| 4 | Qwen 3.7 Max | 68.99 |
| 5 | Gemini 3.5 Flash | 67.09 |
| 6 | Seed 2.1 Pro | 67.09 |
| 7 | Kimi K2.6 | 65.74 |
| 8 | O3 Pro | 64.63 |
| 9 | Kimi K3 | 64.23 |
| 10 | GPT-5.1 | 64.08 |
| 11 | DeepSeek V4 Pro (Max) | 64.08 |
| 12 | Gemini 3.1 Pro (Preview) | 64 |
| 13 | Claude Opus 4.8 | 63.52 |
| 14 | Seed 2.0 Pro | 63.52 |
| 15 | GPT-5 (High) | 63.28 |
Interactive version: theaggregate.ai/benchmark?slug=agi-eval-community-instruction-following · How It Works · Data refreshed daily, snapshot 2026-09-19.