Command A
cohere/command-aThe enterprise generation model, built for RAG, citations and private deployment.
256K ctx · Price on request
Cohere
Enterprise retrieval and language stack
Models
3
Founded
2019
Country
Canada
Toronto
Cohere has been building models out of Toronto, Canada, since 2019. Webparam tracks three of its models, spanning language, embeddings and search: free to call and reaching 256K.
Models
The Cohere catalogue
Ordered by weekly usage on Webparam.
Command A by Cohere CohereOn request Embed v4 by Cohere CohereOn request Embed v4
cohere/embed-v4Multimodal enterprise embeddings: text and images in one vector space, 128K inputs.
128K ctx · Price on request
Rerank 3.5 by Cohere CohereOn request Rerank 3.5
cohere/rerank-3.5The retrieval quality multiplier, reorders candidate documents by true relevance, per query.
4K ctx · Price on request
Recent releases
What shipped lately
History
Cohere never chased consumer chat. It built the enterprise retrieval stack instead. Command handles generation, Embed and Rerank power search pipelines, and private-deployment options made it a fixture in regulated industries where data cannot leave the building.
Cohere never chased consumer chat. It built the enterprise retrieval stack instead. Command handles generation, Embed and Rerank power search pipelines, and private-deployment options made it a fixture in regulated industries where data cannot leave the building.
Strengths
- Complete RAG stack: generate, embed, rerank
- Private and on-prem deployment options
- Enterprise-grade multilingual support
Use Cohere models through Webparam
Every model above is callable through one OpenAI-compatible convention: change the model ID, keep the request.
Related providers