Join waitlist
Back to products
Tiny Aya

Tiny Aya

Local, open-weight AI designed for real-world languages

Website cohere.com
Overview

What it is

We build high performance, secure language models for the enterprise. Our customizable, high-performance language models work on public, private, or hybrid clouds to ensure data security & exceptional support.

Intent

I need it when

Reduce API costs while maintaining multilingual support for production applications

Aya Expanse models (8B and 32B) offer efficient token-based pricing at $0.50-$1.50 per million tokens, providing cost-effective multilingual inference compared to larger proprietary models

Access open-source multilingual models for research and experimentation without licensing restrictions

Aya models are available as open-source research models, allowing researchers and developers to download, fine-tune, and experiment with multilingual capabilities on their own infrastructure

Build multilingual AI applications that serve global audiences across 70+ languages

Aya is a family of multilingual research models covering 70+ languages, enabling developers to create applications that understand and generate text across diverse linguistic markets without building separate language-specific models

Integrate multilingual language understanding into existing enterprise systems with private deployment

Aya models can be deployed privately within Model Vault or on-premises infrastructure, allowing enterprises to maintain data sovereignty while leveraging multilingual AI capabilities

Drop

Not a fit when

  • User requires real-time speech-to-text capabilities; Aya is a text-based multilingual model, not a speech recognition system
  • User needs enterprise workplace automation with agentic workflows; North platform is better suited for this use case
  • User requires semantic search and document ranking; Embed and Rerank models are purpose-built for retrieval tasks
  • User needs production deployment with guaranteed SLA and dedicated infrastructure; Model Vault provides managed deployment but requires separate pricing negotiation
  • User operates in a single language only; Aya's value proposition centers on 70+ language support, making it overengineered for monolingual use cases
Commercials

Pricing

Pay-as-you-go token-based pricing for API access. Aya Expanse models (8B and 32B) charged at $0.50/1M input tokens and $1.50/1M output tokens. Trial API keys are free but rate-limited and not for production use. View pricing