ViMU - Structured Subtext Understanding: leaderboard

ViMU (2026) benchmarks video metaphorical understanding: whether models grasp the implicit, ironic or socially coded meaning of short online videos. This board is the mean in percent of the rhetoric mechanism and social value signal identification scores; higher is better.

Metric: Structured subtext understanding (%): the mean of the rhetoric mechanism and social value signal identification scores on ViMU, short online videos whose meaning lies in subtext (irony, mockery, criticism), uniformly sampled frames, zero-shot through official implementations or APIs; questions and references written by GPT-5.4 and reviewed by human experts; higher is better. Source: arxiv.org. Saturation forecast: Around 2029. 16 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Grok 4.1 Fast31.82#427
2O4 Mini31.36#368
3Gemini 3 Flash (Preview)30.94#152
4Qwen 3.5 27B30.29#375
5Qwen 3 VL 32B Instruct21.41#606
6Ministral 8B21.16#4857
7Gemma 3 27B (IT)20.21#1324
8MiMo-V2-Omni19.78#465
9GPT-5.218.85#185
10Seed 2.0 Lite17.74#300
11Ministral 3 14B16.93#1870
12Gemma 3 4B (IT)14.13#4139
13GLM-4.5V9.06#754
14GPT-5.4 Mini7.97#406
15GPT-4.1 Nano5.67#2056

No result here: #1 Claude Fable 5.2, #3 Gemini 4 Neon, #4 GPT-6.2 Astra.

Interactive version: theaggregate.ai/benchmark?slug=vimu-structured-subtext-understanding · How It Works · Data refreshed daily, snapshot 2026-11-21.