EmoTrans - Emotion State Identification: leaderboard
EmoTrans (2026) emotion state identification: the model names each target person's emotion before and after a transition in a real-world video clip; accuracy (%) over 724 questions, higher is better.
Metric: Accuracy (%) on Emotion State Identification (724 questions: each person's emotion before and after the change), on EmoTrans, built from 1,000 manually annotated real-world video clips (OpenHumanVid and CH-SIMS v2) over 12 scenarios with one or more target persons; zero-shot, video with its text (and audio where the model accepts it); higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 15 models tracked. Updated 7 Oct 2026.
Top models
| # | Model | Setting | Score | Overall rank |
|---|---|---|---|---|
| 1 | Seed 2.0 Lite | 41.2 | #300 | |
| 2 | GPT-5.4 | 40.5 | #135 | |
| 3 | Qwen 3.5 122B A10B | 38.8 | #346 | |
| 4 | Qwen 3.5 27B | 38.4 | #375 | |
| 5 | Gemini 3 Flash | Non-reasoning | 38.3 | #175 |
| 6 | GLM-4.5V | 38 | #754 | |
| 7 | GLM-4.6V | 37.4 | #710 | |
| 8 | Gemini 3 Pro | 36.5 | #145 | |
| 9 | Qwen 2.5 VL 72B Instruct | 33.1 | #841 | |
| 10 | Qwen 3 VL 32B Instruct | 31.2 | #606 | |
| 11 | InternVL3.5-14B | 30.8 | #1504 | |
| 12 | Gemini 2.5 Flash | Non-reasoning | 26.8 | #463 |
| 13 | Qwen 3 VL 8B | 24.4 | #917 | |
| 14 | Qwen 3 Omni 30B A3B | 23.8 | #847 | |
| 15 | InternVL3.5-8B | 21.8 | #1604 |
No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.
Interactive version: theaggregate.ai/benchmark?slug=emotrans-emotion-state-identification · How It Works · Data refreshed daily, snapshot 2026-11-21.