GapEval - Instruction Following (Understanding): leaderboard
Metric: Accuracy (%; describe the result of explicit and implicit image-region edits; text answers to GapEval's bidirectional questions, judged correct or incorrect by GPT-5-mini against reference answers, mean of ten sampled runs). Source: arxiv.org. Saturation forecast: Estimated already saturated. 15 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | GPT-5 | 95.45 |
| 2 | GPT-5 Mini | 93.6 |
| 3 | Gemini 2.5 Flash | 85.11 |
| 4 | Qwen 3 VL 235B A22B | 85.11 |
| 5 | Qwen 3 VL 8B | 82.97 |
Interactive version: theaggregate.ai/benchmark?slug=gapeval-instruction-following-understanding · How It Works · Data refreshed daily, snapshot 2026-09-26.