# Cohere — Multilingual Bridges: How Data Mixing Unlocks In-Language Reasoning

- Company: Cohere (cohere.com)
- Announced: 2026-10-06
- Category: research-paper
- Coverage: not counted
- Announcement: yes
- Group: announcements
- Source: https://cohere.com/blog/building-multilingual-bridges
- Record: https://forck.live/items/16191-multilingual-bridges-how-data-mixing-unlocks-in-language-reasoning
- Subject: Command
- Open weights: yes
- Models affected: Tiny Aya L2-Thinker, Qwen3.5-4B
- Context window: 32K context

Cohere's research shows that multilingual models can be trained to reason in the user's language, not just answer in it, through a data mixing strategy built on three pillars: broader language coverage, a small share of multilingual non-reasoning data, and English reasoning data as a backbone. The resulting model, Tiny Aya L2-Thinker (3.35B), reasons in the prompt language over 93% of the time across 60 languages with minimal accuracy loss, and the weights and data are released. The research also finds that inference-time language forcing approaches are ineffective, so in-language reasoning must come from training.

## Evidence

Verbatim from https://cohere.com/blog/building-multilingual-bridges:

> Our resulting model, Tiny Aya L2-Thinker (3.35B), reasons in the prompt language over 93% of the time across 60 languages. It does so with minimal accuracy loss, holds up on low-resource languages, and uses fewer reasoning tokens than comparable models.

---

Record: https://forck.live/items/16191-multilingual-bridges-how-data-mixing-unlocks-in-language-reasoning
Catalogue: https://forck.live/llms.txt
Current issue: https://forck.live/feed.md