Source: https://cohere.com/chat and related Cohere docs pages (chat API docs, models docs, changelog, pricing). Summary of findings: Cohere positions its Chat offering as an enterprise-focused, privacy-first conversational capability that can be deployed in private VPCs, on-premises (air-gapped), or via dedicated deployments to meet data residency and regulatory requirements. The product emphasizes multi-layered security and compliance controls, model customization and training on customers' proprietary data, and integration with a broader enterprise product ecosystem (mentions North to unify tools/workflows and other products such as Compass to provide search, agents, and knowledge workflows). Model capabilities highlighted include high-performance generative models, semantic text representations (embeddings / vector search) for document comparison and retrieval, and relevance/result refinement (reranking) models to optimize and personalize search results. The Chat API supports streaming (token-by-token via SSE), role tokens (user/assistant/system/tool), tools/function calls, and standard generation controls (temperature, top_p, frequency/presence penalties, etc.). Authentication is via Bearer tokens. Developer readiness: API access, deep technical docs, and a Playground are available. The page notes support for multiple languages (23 languages referenced). CTAs link to documentation, Playground, Pricing, and Contact Sales / Request a demo. Pricing summary from the pricing page: a free trial API key is available at signup (rate-limited, not for production/commercial use), production access is billed by token usage (apply for a Production API key via dashboard; invoices monthly or when $250 outstanding), enterprise product pricing (North, Compass, dedicated deployments) is custom and requires contacting sales/requesting a demo. The pricing page also lists legacy per-model token rates for existing customers; examples include Command (legacy) and other legacy/variant rates, plus research-model examples (e.g., Aya Expanse example rates). Support/contact pointers: contact-sales page, docs/frequently asked questions, and [email protected] as listed in docs/FAQs. Other facts: Cohere’s Chat API with retrieval/RAG capabilities had a public beta/launch announcement in September 2023 (press coverage referenced), and the docs changelog shows ongoing model updates and evolving model names and variants. All details in this record are taken directly from the visited pages and linked documentation; no additional claims or unverifiable data were added.