Cohere

Cohere provides model APIs for chat, embeddings and reranking. Its Rerank API takes a question and a list of texts and returns them ordered by relevance with a score for each.

Website Docs Changelog Compare with others

Cohere at a glance

From official pages, checked Oct 11, 2026. Tap the arrow for each source.

Free tier: Yes
Trial API key calls are free and rate limited. Trial keys are not allowed for production or commercial use.
Rate limit: 20 chat requests per minute
Trial key: 20 chat requests a minute, 1,000 API calls a month. Production Command A chat: 500 a minute.
Size limit: 128k token context
Command A+: 128k context, 64k max output tokens. Command A: 256k context, 8k max output tokens.
Official SDKs: PythonJavaScript / TypeScript
Trains on your data: Opt-out
Opt out of training on prompts and generations under Data Controls in Settings; applies to new data only.
EU hosting: No
Trust Center: SaaS runs on Google Cloud US-Central with no servers outside the US. Private deployment is offered.
Compliance: SOC 2ISO 27001GDPR
SOC 2 Type II audit and ISO 27001 certificate; GDPR handled through a data processing addendum.

What you can build with Cohere

CohereRecipeConfig

Reorder search results by relevance to a question with Cohere Rerank

Send a question and a list of text passages to Cohere Rerank from Python and get them back ordered by relevance, each with a score.

Free tier: Yes

Cohere at a glance

What you can build with Cohere

Reorder search results by relevance to a question with Cohere Rerank

Recent updatesAll updates