North Small Translate delivers fast high-quality machine translation in 50 languages
North Small Translate: Advanced Cost-Effective Translation (Cohere CAT+)
Computation and Language
Summary
Translation software helps people understand text written in different languages, but it can be slow or inaccurate. The authors created North Small Translate, a large AI model designed to translate up to 50 languages quickly and well. It uses a smart training method that focuses on hard examples and a mix of learning steps to improve. Unlike some translation tools, it doesn’t need slow reasoning during use, making it fast. It also supports related tasks like improving translations and checking their quality.
What this means in practice
- •For language service providers: Integrate fast, high-quality machine translation across 50 languages into client workflows without expensive inference costs.$Commercial implications: Enables scalable and cost-effective multilingual translation services that outperform other sub-1T parameter models.
- •For software localization teams: Use the model’s post-editing and quality estimation capabilities to automate translation quality checks and improve localized software efficiently.
Authors
Tom Kocmi, Alexandre Bérard, Phil Blunsom, Samuel Cahyawijaya, Shaun Cassini, Nicholas Frosst, Ona de Gibert, Aidan Gomez, Nithya Govindarajan, Shun Kiyono, Olivia Lasche, Lawrence Rogers, Kelly Marchisio, Nikita Moghe, Yash More, Camila Moran-Hidalgo, Yiyang Nan, Michael Sachs, Trisha Starostina, Daan van Stigt, Spencer Rarrick, Sebastian Vincent, Ivan Zhang
Abstract
We present North Small Translate, an open-weight, LLM-based machine translation (MT) model with instruction-following capabilities built on the same foundation as Cohere's Command A Plus, a mixture-of-experts architecture with 25 billion active parameters out of 218 billion total parameters. North Small Translate is trained using difficulty sampling to obtain challenging documents and a five-step training protocol combining supervised fine-tuning, direct preference optimization, and online reinforcement learning. We prioritized throughput through a non-reasoning base model and supplemented with optional agentic capabilities to unlock translation quality gains. North Small Translate is trained to perform MT-related tasks, including post-editing and quality estimation, as well as related tasks such as general instruction following. The model achieves top MT performance across 50 languages in the class of models under 1T parameters, with no need to run expensive reasoning at inference time.