AICA-Bench: leaderboard
Metric: Overall average on a percent scale: unweighted mean of the emotion understanding average (weighted F1 of basic and CoT prompting), the emotion reasoning score and the emotion-guided content generation score; reasoning and generation are 1-5 ratings rescaled to 20-100, so the floor is 40/3, over AICA-Bench (8,086 affective images from nine public emotion datasets with GPT-4o-generated instructions; closed-source models through their APIs at standard settings, open models on A100 GPUs); higher is better. Source: arxiv.org. Saturation forecast: Around January 2027. 23 models tracked.
Top models
| # | Model | Score |
|---|---|---|
| 1 | Gemini 2.5 Pro | 73.49 |
| 2 | GPT-4o | 72.82 |
| 3 | Gemini 2.5 Flash | 71.14 |
| 4 | GPT-4o Mini | 70.81 |
| 5 | Gemini 2.0 Flash | 67.68 |
| 6 | Qwen 2.5 VL 7B Instruct | 65.78 |
| 7 | Qwen 2 VL 7B Instruct | 61.45 |
| 8 | MiniCPM-V-2.6 | 58.08 |
Interactive version: theaggregate.ai/benchmark?slug=aica-bench · How It Works · Data refreshed daily, snapshot 2026-10-07.