ai api
Cohere API
Cohere provides hosted language and retrieval model APIs, including generation, embeddings, reranking, transcription, and document parsing. API pricing is model- and task-dependent, while enterprise deployments are also available.
Data status: partially_verified · Checked: 2026-10-16
Good fit if
Generation, embedding, reranking, transcription, and parsing APIs
May not fit if
Trial API keys are rate-limited and are not permitted for production or commercial use.
Features and benefits
- Generation, embedding, reranking, transcription, and parsing APIs
- Model-specific token or task-unit billing for production usage
- Trial API keys are available for evaluation with published rate limits
Limitations
- Trial API keys are rate-limited and are not permitted for production or commercial use.
- Production pricing depends on the selected model and billed token or task units; no single API price represents the catalog.
- Some newer models require contacting sales for production access or limits.
Pricing
Trial API
0 USD · free
Free evaluation API key; rate-limited, not permitted for production or commercial use. Model and endpoint limits apply.
Pricing sourceProduction API
Price unknown · usage_based
Model-specific pricing. Generation is billed per input/output token, reranking per search unit, and embedding by tokens; see current model rates on the official pricing page.
Pricing sourceModel Vault Embed 5 Fast Small
3 USD · usage_based
Published rate per instance-hour; the same page lists a $2,000 monthly rate per instance. Applies to managed Model Vault deployment, not general API inference.
Pricing sourceSources
- https://cohere.com/ · 2026-10-16: official product overview
- https://docs.cohere.com/v2/docs/how-does-cohere-pricing-work · 2026-10-16: pricing model, trial API terms, production API billing units
- https://docs.cohere.com/docs/rate-limits · 2026-10-16: trial and production rate limits
- https://cohere.com/pricing · 2026-10-16: Model Vault pricing, hourly and monthly instance rates