Visual Aesthetic Benchmark
product→ stable
VAB
Visual Aesthetic Benchmark is a test that reveals frontier multimodal large language models judge aesthetics correctly only 26.5% of the time.
1Total Mentions
+0.10Sentiment (Neutral)
0.0%Velocity (7d)
First seen: May 14, 2026Last active: May 14, 2026
Signal Radar
Five-axis snapshot of this entity's footprint
Loading radar…
Mentions × Lab Attention
Weekly mentions (solid) and average article relevance (dotted)
mentionsrelevance
Loading timeline…
Timeline
1- Product LaunchMay 14, 2026
VAB benchmark released showing MLLMs achieve 26.5% accuracy vs 68.9% for humans
View source- accuracy:
- 26.5%
- human accuracy:
- 68.9%
- tasks:
- 400
- images:
- 1195
Relationships
No relationships mapped yet.
Recent Articles
No articles found for this entity.
Predictions
No predictions linked to this entity.
AI Discoveries
No AI agent discoveries for this entity.
Sentiment History
Positive sentiment
Negative sentiment
Range: -1 to +1
| Week | Avg Sentiment | Mentions |
|---|---|---|
| 2026-W20 | 0.10 | 1 |