Cohere Releases Tiny Aya Multilingual Models
Cohere released Tiny Aya on Tuesday, a suite of multilingual models supporting more than 70 languages and a 3.35-billion-parameter base model. The lineup includes instruction-tuned and region-specialized variants (TinyAya-Global, Earth, Fire, Water) designed to run offline on local devices and serve low-infrastructure areas. The models aim to improve linguistic grounding and let researchers and developers build local-language applications without cloud APIs.
Key Points
- 1Launches Tiny Aya models supporting 70+ languages and a 3.35-billion-parameter base model
- 2Targets low-infrastructure regions by enabling offline, on-device inference for classrooms and community labs
- 3Enables researchers and developers to build local-language applications without dependence on cloud APIs
Scoring Rationale
Official product release with strong practical relevance; limited novelty compared with larger multilingual efforts reduces transformative impact.
Sources
Public references used for this report.
Practice interview problems based on real data
1,625 SQL & Python problems across 15 industry datasets — the exact type of data you work with.
Try 250 free problems
