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

#ModelScore
1GPT-595.45
2GPT-5 Mini93.6
3Gemini 2.5 Flash85.11
4Qwen 3 VL 235B A22B85.11
5Qwen 3 VL 8B82.97

Interactive version: theaggregate.ai/benchmark?slug=gapeval-instruction-following-understanding · How It Works · Data refreshed daily, snapshot 2026-09-26.