Cohere Releases North Small Translate: A 218B MoE Translation Model That Scores 83.6 on WMT26 Across 50 Languages

Cohere has released North Small Translate, an open-weight Mixture-of-Experts model built for machine translation across 50 languages.

Model Architecture and Scale

The model uses 25B of its 218B parameters per token, reflecting its Mixture-of-Experts design.

Performance

North Small Translate scores 83.6 on Cohere’s WMT26 evaluation.

Availability and Licensing

• Weights are free for non-commercial use.
• Commercial access is available through Cohere Model Vault or RWS Language Weaver.