Skip to content
In development

Evaluation platform and leaderboard family

OpenCompass

A broad evaluation ecosystem whose different leaderboards need their own context.

Key information

Focus
Language and multimodal model evaluation across distinct benchmark collections.
Measure
Leaderboard-specific objective and subjective measures.
Version / scope
CompassRank / CompassBench; each leaderboard has its own protocol
Update pattern
CompassBench rules describe quarterly benchmark updates. That does not imply every OpenCompass project follows the same calendar.
Publisher update
Not consistently disclosed
Editorial verification
15 October 2026

In plain English

What makes it different?

OpenCompass provides evaluation software and a family of leaderboards. Its breadth makes it useful for discovering evaluations, but the family name is not one single benchmark. Some rankings use closed benchmark collections; individual multimodal projects have their own lifecycles.

Original scor.es editorial guide.
Sources and attribution are listed below.

What it measures

Language and multimodal model evaluation across distinct benchmark collections.

  • Intelligence
  • Reasoning
  • Coding
  • Multimodal

Strengths

  • Broad coverage of model types and evaluation areas.
  • Open evaluation tooling and public ranking rules provide useful context.

Limitations

  • Different collections and subjective judges produce different measures.
  • Some evaluation sets are closed. MMBench's official page says its online service and leaderboard updates stopped on 31 March 2026.

How the evaluation works

The platform supports evaluations across selected datasets and model configurations. Public ranking rules distinguish objective from subjective assessment.

CompassBench describes quarterly benchmark changes. Other projects, datasets and leaderboards must be checked separately before treating their scores as comparable.

Configuration matters

Identify the exact leaderboard, dataset collection, language, prompt, judge and model revision. Do not present an archived subproject or a software release as a refreshed CompassRank result.

Primary methods: Official leaderboard rules ↗MMBench service-closure notice ↗

Results at the publisher

Scores stay with the original source

We have not documented the permissions needed to reproduce this leaderboard here. You can still explore its methods and visit the publisher for results.

Open OpenCompass ↗

Versions & freshness

Verified 15 October 2026

No single current result version was confirmed for the whole ecosystem. The main portal is JavaScript-rendered; rules and repository documentation were independently readable.

Official portals available as JavaScript applications; repository, ranking rules and MMBench status readable. CompassBench rules describe quarterly benchmark updates. That does not imply every OpenCompass project follows the same calendar.

Our review date is not an evaluation date. Individual run dates are not consistently available; they remain unknown here. No automatic result collection is enabled.

Publication & reuse

Permission not documented

OpenCompass software is Apache 2.0. The constituent datasets, closed benchmark collections and hosted leaderboard results do not automatically share that licence.

Publication, data retention and commercial reuse require a decision for the actual leaderboard and fields, not just the platform licence.

View the scope of the reuse decision
Result fields
No result fields cleared
Public result display
not enabled
Storage
not enabled
Historical copies
not enabled
Updates
not enabled
Commercial reuse
not enabled
Derived ratings
No scoped grant documented; not enabled.
Attribution
Credit the named publisher and link the primary sources. Attribution alone does not grant reproduction rights.

This decision does not clear mixed-source scientific recommendations. Publisher content, software, datasets and model licences are separate. No partnership is implied.

Primary sources & attribution

This guide is original scor.es commentary about OpenCompass team / Shanghai AI Laboratory's work. The links below support the identity, methods, versions and reuse review.

  1. Official portal ↗Checked 15 October 2026
  2. Official CompassRank portal ↗Checked 15 October 2026
  3. Official leaderboard rules ↗Checked 15 October 2026
  4. Official evaluation repository ↗Checked 15 October 2026
  5. MMBench service-closure notice ↗Checked 15 October 2026

Editorial edition 2026-10-15.1 · reuse review directory-rights-1. Re-check due 14 November 2026.