OmniCap-IF - Format: leaderboard

Metric: Format constraint satisfaction rate (%) (length, JSON schema, ordered lists and other structural or stylistic constraints) on multi-constraint captioning instructions on videos with sound; format constraints are checked by rule-based tools on LLM-extracted structure, content constraints by temporal-IoU tools or a gpt-5-mini judge answering binary and multiple-choice questions against the caption; higher is better. Source: arxiv.org. Saturation forecast: Around December 2026. 14 models tracked.

Top models

#ModelScore
1Gemini 3.1 Pro (Preview)90.45
2Gemini 3 Flash88.57
3MiMo-V2.586.4
4MiMo-V2-Omni80.6
5MiniCPM-o-4.578.6
6Qwen3 Omni 30B A3B Instruct77.37
7Qwen2.5-Omni-7B62.97

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