Cohere
Cohere is an AI company focused on enterprise use cases, offering large language models (the Command family, currently led by Command A and Command A+), embedding models, and rerank models for search and retrieval, through an API billed per token or per dedicated instance. Cohere offers free evaluation API keys with limited usage, and production access is paid, with pricing for its newest flagship models generally quoted directly by its sales team.
Facts checked on against the sources listed at the end of this page.
- Pricing
- Pay as you goPaid from $0.30 input / $0.60 output per million tokens on Command-light; current flagship Command A models are priced on request
- Open source
- NoProprietary
- npm downloads a week
- 543kdown 2% on the week before
A good fit for
- Enterprise search and retrieval-augmented generation applications
- Teams that want embedding and rerank models alongside a language model from one provider
- Organizations already billed through Amazon Bedrock, Google Cloud or Azure that want Cohere models on the same bill
- Workloads that benefit from dedicated, reserved model capacity through Model Vault
Not a good fit for
- Developers who want fully transparent, published pricing for every current model without contacting sales, since flagship Command A pricing isn't listed publicly
- Consumer-facing chat products, since Cohere is positioned mainly toward enterprise and developer use cases
- Teams wanting an open-weight model they can self-host, since Cohere's models are proprietary and API-only
Cohere pricing
Evaluation API keys are free but limited in usage. Older Command and Command R models are billed per million tokens, for example $0.30 input / $0.60 output on Command-light up to $3.00 input / $15.00 output on Command R+. Pricing for the current Command A and Command A+ flagship models is quoted directly by Cohere's sales team, and Model Vault dedicated instances are billed hourly or monthly.
| Plan | Price | What you get |
|---|---|---|
| Evaluation key | $0 | Free but limited usage for testing the API. |
| Command-light | $0.30 input / $0.60 output per million tokens | Cohere's smallest, lowest-cost language model. |
| Command R+ (04-2024) | $3.00 input / $15.00 output per million tokens | A larger, more capable Command R generation model. |
| Command A / Command A+ | Contact sales | Cohere's current flagship models. |
| Model Vault (Embed 4, Rerank, Parse) | From $4 an hour or $2,500 a month | Dedicated, reserved model instances rather than per-token billing. |
Prices in USD, checked on 27 September 2026. Current prices on cohere.com.
Main features
- Command language models
- Cohere's LLM family, currently led by Command A and Command A+, plus older Command R models.
- Embed models
- Turn text into vector embeddings for search and retrieval applications.
- Rerank models
- Reorder search results by relevance, commonly used alongside a vector search step.
- Model Vault
- Dedicated, reserved model instances billed hourly or monthly instead of per token.
- Multi-cloud availability
- Models accessible through Amazon Bedrock, Google Cloud and Microsoft Azure as well as Cohere's own API.
- Evaluation API keys
- Free, usage-limited keys for testing before moving to a paid production key.
About Cohere
Cohere builds large language models under the Command name, alongside Embed models for turning text into vector representations and Rerank models for improving search result ordering, aimed primarily at enterprise search, retrieval-augmented generation and internal tooling use cases.
Its current flagship models are Command A and Command A+, offered alongside older Command and Command R family models that remain available at published per-token prices. Cohere also offers Model Vault, dedicated model instances billed hourly or monthly rather than per token, for customers who want reserved capacity.
Cohere's models are available directly through its own API and also through major clouds including Amazon Bedrock, Google Cloud and Microsoft Azure, letting enterprise customers billed through those platforms use Cohere's models without a separate account.
- Made by
- Cohere
- Free plan
- Yes: evaluation API keys are free but limited in usage
- Available on
- REST API, Amazon Bedrock, Google Cloud, Microsoft Azure
Usage and activity
Weekly npm downloads of cohere-ai, last 12 months.
Cohere alternatives
Cohere compared
CoherevsMistral AI
Cohere builds proprietary language, embedding and rerank models aimed at enterprise search and retrieval, with its current flagship pricing quoted directly by sales.
CoherevsOpenAI API
Cohere is an enterprise-focused AI company offering its Command language models plus embedding and rerank models for search and retrieval, available through its own API and major clouds.
Questions about Cohere
Is Cohere free to use?
Evaluation API keys are free but limited in usage. Production use is paid, billed per token for older Command models or by contacting sales for current Command A models.
What is Cohere's current flagship model?
Command A and Command A+ are Cohere's current flagship language models, alongside the older Command R family that remains available at published prices.
Does Cohere offer embeddings and search tools?
Yes, Cohere offers Embed models for vector embeddings and Rerank models for improving search result relevance, alongside its language models.
Can I use Cohere through AWS or Azure?
Yes, Cohere's models are available through Amazon Bedrock, Google Cloud and Microsoft Azure, in addition to Cohere's own API.
Sources and links
Work on Cohere? Claim this listing. Something wrong or out of date? Suggest a correction. Stats come from the public GitHub and npm APIs and update every day.