AI Language Benchmarks | Tracking language proficiency of AI models for every language

About this tool

languagebench provides AI model evaluations for every language in the world.

Who is this for?

• Practitioners can pick the best model for a given language.
• Policymakers and funders can identify and prioritize neglected languages.
• Model developers can compete on our benchmarks.

⚡ Live Updates

Benchmark results automatically refresh every night and include the most popular models from OpenRouter, plus community-submitted models.

Authors

languagebench is a collaboration between BMZ’s Data Lab, the BMZ-Initiative Fair Forward (implemented by GIZ), and the E&E group of DFKI’s Multilinguality and Language Technology Lab.