Benchmark Explorer

Explore how models perform on various benchmarks

Benchmarks

🔍

Vals Multimodal Index

A composite benchmark measuring multimodal understanding across vals.ai's enterprise benchmarks including vision-language tasks. This index extends the standard Vals Index by incorporating evaluations that require processing and reasoning about visual information alongside text, testing models on real-world multimodal enterprise scenarios.
Source:

Model Performance

#1
GPT-5.2
63.9%
#2
Claude Opus 4.5 (Thinking)
63.4%
#3
Claude Opus 4.5
61.4%
#4
Gemini 2.5 Pro (Jun 2025)
51.3%
#5
Grok 4.20 (Reasoning)
46.3%
#6
Grok 4.20
36.0%