Cultural Moment Benchmark - Concept Naming: leaderboard
The Cultural Moment Benchmark (CMB, 2026) evaluates video cultural reasoning on 306 expert-curated concepts from seven Southeast Asian countries in five categories, in three stages per concept. This board is Stage 1: choosing the concept name that a symbolic description refers to from four candidates with semantically similar distractors. Accuracy in percent; higher is better.
Metric: Accuracy (%; Stage 1, four-option choice of the concept name from a symbolic description, distractors by semantic similarity; 306 expert-curated concepts from seven Southeast Asian countries in five categories (celebrations, dances, games, music, weddings); Carry mode, where the model's own previous-stage answer is carried into the next stage; Gemini, GPT and Qwen read video at 1 fps, InternVL at its native 12 frames). Source: arxiv.org. Saturation forecast: Around February 2027. 6 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Gemini 3.1 Pro (Preview) | 77.8 | #91 |
| 2 | GPT-5.4 | 70.9 | #121 |
| 3 | Qwen 3.5 27B | 53.6 | #358 |
| 4 | Qwen 3 VL 32B Instruct | 51 | #582 |
No result here: #1 Claude Fable 5.2, #2 Gemini 4 Carbon, #5 Claude Opus 5.5.
Interactive version: theaggregate.ai/benchmark?slug=cultural-moment-benchmark-concept-naming · How It Works · Data refreshed daily, snapshot 2026-11-12.