Cohere Embed AI features and pricing (October 2026)

Cohere Embed is an embedding model family from Toronto-based Cohere. It turns text, images and mixed documents into vectors for semantic search, retrieval-augmented generation and agents in companies that need multilingual retrieval.

Embed is part of Cohere's enterprise platform, next to the Command language models, Rerank and the North workplace product. Embed 4 accepts text, images and mixed files such as PDFs, supports a context of 128,000 tokens and returns vectors of 256 to 1,536 dimensions. Customers use the hosted API, dedicated Model Vault instances, or private deployments on premises or in isolated VPCs. According to Cohere's trust center, the company holds SOC 2 Type II, ISO/IEC 27001 and ISO/IEC 42001.

• Finds documents across languagesEmbed places queries and documents in one vector space regardless of language, covering more than 100 languages. A German query can retrieve an English contract, so international teams search one index instead of one per language.
• Embeds scanned PDFs and slidesEmbed 4 creates a single vector for pages that mix text and images. Knowledge teams can index reports and slide decks without first converting every page to clean text.
• Reduces vector storageOutput dimensions can be set to 256, 512, 1,024 or 1,536. Engineering teams trade a little accuracy for smaller indexes and lower database costs where that fits.

• Private deployment on premises or in isolated VPCs next to the hosted API.
• One vector for mixed text and image pages with a 128,000-token context.
• Headquartered in Toronto, with offices in Europe and Asia.

Cohere was founded in Toronto in 2019 and is based there, with further offices in New York, London and Paris among others.

Decide between the hosted API, Model Vault and private deployment based on where your documents may be processed. Trial keys are free but rate limited and not meant for production. Production keys are billed by usage, and token prices for Embed 4 were not listed on the pricing page at our check, so request them before you plan volumes. Plan for re-embedding your corpus if you switch embedding models later, since vectors from different models are not compatible.

No reviews yet. Be the first to share how your team uses it.

Every review shows its verification level. We read each one before it goes live. How we check reviews