ViMU - Rhetoric Mechanism Identification: leaderboard

ViMU (2026) benchmarks video metaphorical understanding: whether models grasp the implicit, ironic or socially coded meaning of short online videos. This board is a multiple-choice score in percent for identifying the rhetorical devices used, with zero credit for any wrong selection and partial credit otherwise; higher is better.

Metric: Rhetoric mechanism identification score (%): selecting the rhetorical devices (irony, exaggeration, juxtaposition and others) that carry the meaning, set-based multiple-choice scoring: 0 if any selected option is wrong, otherwise the share of gold options selected, averaged over items, 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 2031. 16 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1Qwen 3.5 27B38.18#375
2Grok 4.1 Fast34.91#427
3Gemini 3 Flash (Preview)33.63#152
4O4 Mini33.21#368
5Gemma 3 27B (IT)32.47#1324
6Ministral 8B31.87#4857
7Qwen 3 VL 32B Instruct27.65#606
8Ministral 3 14B27.29#1870
9Gemma 3 4B (IT)21.1#4139
10MiMo-V2-Omni21.04#465
11Seed 2.0 Lite18.75#300
12GPT-5.216.55#185
13GLM-4.5V8.87#754
14GPT-5.4 Mini4.17#406
15Claude 3 Haiku2.99#2663

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

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