InsightVQA - Emotion Perception: leaderboard

Metric: Accuracy (%) on single-choice emotion-category and valence recognition questions (the answer plus three LLM-written distractors) in InsightVQA-Bench, the held-out test split (a tenth of the data, about 30,000 questions) of emotion-annotated images from six public sources; higher is better. Source: arxiv.org. Saturation forecast: Around 2030. 14 models tracked.

Top models

#ModelScore
1Qwen 2.5 VL 7B Instruct57.95
2Gemini 2.5 Flash56.83
3Claude 3.7 Sonnet56.37
4Qwen 2.5 VL 72B Instruct53.83
5InternVL3.5-8B53.4
6Qwen 2.5 VL 32B Instruct52.8
7DeepSeek V3.251.2
8GPT-4o50.63

Interactive version: theaggregate.ai/benchmark?slug=insightvqa-emotion-perception · How It Works · Data refreshed daily, snapshot 2026-09-29.