Cohere
Builds enterprise-focused large language models accessible via API for text generation, retrieval, and classification.
Updated April 2026
Overview
- Website
- cohere.com
- Founded
- 2019
- Headquarters
- Toronto, Canada
- Segment
- Frontier Foundation Model Labs
- Posture
- Enterprise LLM
Product overview
Cohere develops frontier LLMs like Command R/R+ (generative), Embed (semantic representation), and Rerank (result optimization), plus platforms like North (AI workspace) and Compass (enterprise search). Used by enterprises including RBC, Fujitsu, LG, Oracle, and Salesforce for secure RAG, automation, and multilingual applications. Distinct from consumer-oriented labs like OpenAI by prioritizing private/on-prem deployments, data security, customization on proprietary data, and business-specific optimizations over general chatbots.
Revenue model
API usage-based pricing per token (e.g. Command R+ $3/1M input, $15/1M output; Embed English $0.10/1M); Model Vault dedicated instances ($2,500-$3,250/month); custom enterprise licensing for North/Compass and bespoke models.
Moat
- Proprietary Technology
- Switching Costs
- Scale Advantages
- Ecosystem Lock-in
Cohere's competitive moat stems from deep expertise in deploying AI for highly regulated, complex enterprise sectors (government, telecom, banking, healthcare) that competitors avoid, combined with a capital-efficient cloud-agnostic model enabling private/air-gapped deployment and proprietary know-how in RAG and agentic AI platforms.