
Cohere
Enterprise AI models and workplace tools for search, generation, retrieval, automation, plus managed or private deployment.
About Cohere
Cohere is an enterprise AI platform for organizations that need language models, retrieval, workplace search, workflow agents, and tighter control over deployment. North combines generative models, intelligent search, and agents in a workplace interface. Compass is built for search and discovery across business data, using connectors, document parsing, and a managed index. Developers can obtain an API key, follow the documentation, and try ideas in a no-code Playground. Model Vault provides fully managed, dedicated instances for Embed and Rerank models, with hourly and monthly prices listed by performance tier. Cohere also supports layered security controls, configurable handling and retention, opting out of model training, and private deployment. It fits organizations ready for enterprise procurement, integration work, and capacity-based costs. Teams looking for a low-cost consumer assistant will likely find the buying process and deployment model heavier than they need.
Best for
Best use cases
Search and analyze information distributed across company systems
Automate employee tasks and complex workflows with AI agents
Deploy dedicated Embed or Rerank capacity with predictable instance rates
Compare before choosing
Compare custom-priced North or Compass with API-first alternatives
Compare Model Vault instance cost with usage-based hosted model pricing
Check connector, deployment, retention, and infrastructure compatibility before procurement
Features
Enterprise knowledge search
Connect business data and use intelligent search, document parsing, managed indexing, and retrieval models to surface relevant internal information.
Task and workflow automation
Use North and its AI agents to automate routine tasks as well as multi-step enterprise workflows through a workplace interface.
Custom generative AI applications
Build business applications with Cohere generative models, APIs, developer documentation, an API key, and a no-code Playground for testing.
Private model deployment
Run Embed and Rerank models in the fully managed Model Vault, or discuss private deployment options for stricter infrastructure and data-control needs.
Pros
- Supports generation, search, embedding, reranking, and agents in one enterprise product family.
- Offers managed and private deployment paths for teams with infrastructure or data-control requirements.
- Publishes concrete Model Vault hourly and monthly instance rates for cost comparison.
Cons
- North and Compass require a sales conversation, so teams cannot estimate their full cost from a public price table.
- Model Vault starts at $2,500 per month per instance, which may be too expensive for small or lightly used workloads.
- Production adoption can require data connectors, indexing, application integration, and enterprise procurement work.
Pricing Plans
North
- Intuitive interface
- Purpose-built generative models
- Intelligent search
- AI agents for routine tasks and complex workflows
Compass
- Pre-built data connectors
- Intelligent search
- Document parsing
- Managed index
Model Vault: Embed 4 Small
- Embed 4 model
- Small performance tier
- Fully managed dedicated deployment
- Guaranteed performance
Model Vault: Embed 4 Medium
- Embed 4 model
- Medium performance tier
- Fully managed dedicated deployment
- Guaranteed performance
Model Vault: Rerank 3.5 Medium
- Rerank 3.5 model
- Medium performance tier
- Fully managed dedicated deployment
- Guaranteed performance
Model Vault: Rerank 4 Fast Medium
- Rerank 4 Fast model
- Medium performance tier
- Fully managed dedicated deployment
- Guaranteed performance
Model Vault: Rerank 4 Pro Medium
- Rerank 4 Pro model
- Medium performance tier
- Fully managed dedicated deployment
- Guaranteed performance
Model Vault: Rerank 4 Pro Large
- Rerank 4 Pro model
- Large performance tier
- Fully managed dedicated deployment
- Guaranteed performance
* Pricing information is for reference only. Please refer to the latest announcement on Cohere's official website.