Cohere Coral

Cohere is an enterprise AI platform built for private, secure, and customizable deployments across cloud, virtual private cloud, on-premise, and air-gapped infrastructure. Founded in 2019 by former Google Brain researchers including Aidan Gomez (co-author of the original Transformer paper), Cohere has raised $1.1B+ from NVIDIA, Salesforce Ventures, and Cisco, positioning itself as the sovereign AI alternative to OpenAI and Anthropic for regulated industries. The product suite spans three enterprise platform products and four model and infrastructure layers. North is the all-in-one AI workplace platform: an integrated environment for AI agents, intelligent search, and task automation built for enterprise-wide deployment with full data isolation. Compass is Cohere's intelligent search and discovery system, surfacing insights across enterprise data sources via pre-built connectors, document parsing, and a managed index. Command is the flagship generative model suite supporting 23 languages, optimized for RAG-grounded responses, and designed to integrate with existing enterprise systems without disruption. Model Vault is Cohere's dedicated managed deployment platform for running models securely with no shared resources or multi-tenancy overhead. Pricing: North and Compass are custom enterprise, contact sales required. Model Vault pricing is publicly listed: Embed 4 Small at $4/hour or $2,500/month, Embed 4 Medium at $5/hour or $3,250/month, Rerank 3.5 Medium at $5/hour or $3,250/month, Rerank 4 Pro Large at $10/hour or $6,500/month. Fixed and Flex billing plans available. Legacy API developer pricing: Command R+ at $3/$15 per million input/output tokens. Compliance: SOC 2 Type II, ISO 27001, GDPR, HIPAA, and FedRAMP-aligned deployments. Cohere has officially published north-mcp-python-sdk on GitHub for MCP-compatible integrations with North. Deployment options cover AWS, Azure, GCP, OCI, and IBM Cloud, plus private VPC, on-premise, and air-gapped environments, addressing data residency and sovereignty requirements that public API providers cannot match. Not the right fit for: teams wanting self-serve access or a free trial without engaging sales, SMBs or mid-market teams unable to meet the $100K+ annual minimum commitment, development teams looking for a single-model API without enterprise implementation overhead, or organisations on tight timelines since enterprise deployments typically take weeks to months to configure fully. For self-serve API access the developer tier is separate from the enterprise platform.