Skip to content

OpenCompass Multi-modal Leaderboard

A live leaderboard for multimodal model scores

Visit official site

What OpenCompass Multi-modal Leaderboard is

The platform is described as a real-time updated service that evaluates and ranks different multimodal models (VLMs). It computes average scores from 8 multimodal benchmarks and provides detailed performance data, while including only open-source VLMs or publicly available APIs.

What you can do with it

  • View real-time rankings and performance data for multimodal models
  • Read detailed reports with scores for each benchmark
  • Add a new model by creating a PR in VLMEvalKit
  • Contact [email protected] for help and support

Application scenarios

  • Researchers compare how different models perform on specific tasks
  • Developers use the data to choose a model for their application scenario
  • Enterprise decision makers shape technology purchase or research plans

Getting started

  • Visit the OpenCompass Multi-modal Leaderboard site
  • Check the current rankings and performance data
  • Click a model of interest for a more detailed performance report
  • To add a new model, go to VLMEvalKit and create a PR

Pros & cons

✓ What we like

  • A new model can be added by creating a PR in VLMEvalKit
  • Help is available through [email protected]

! What to watch out for

  • Only open-source VLMs or publicly available APIs are included
  • Average scores come from the 8 listed multimodal benchmarks

FAQ

How are average scores computed?

The platform uses 8 multimodal benchmarks to compute the average score.

Which models are included?

Only open-source VLMs or publicly available APIs are included.

Last reviewed: 2026-11-15

More AI office assistant tools

View all →

How we review