DEV INFRASTRUCTURE

A Canadian enterprise AI company that builds the Command family of generative models, Embed models for semantic search and Rerank models for retrieval. Its North platform packages these into an agent workspace for staff. Cohere focuses on private deployment, letting organisations run models in their own cloud, on-premises or fully disconnected environments.

SPEC SHEET

PRICE
Freemium
SET-UP
Advanced
DATA
Can run onshore
FOUNDED
2019
HQ
Toronto, Canada
WEBSITE
cohere.com
REVIEWED
October 2026

001 / WHAT IT DOES

  1. 01Command models for generation, tool use and agents
  2. 02Embed models for multilingual semantic search
  3. 03Rerank models that reorder search results by relevance
  4. 04North agent platform for staff with search and automation
  5. 05Deployment in private cloud, on-premises or air-gapped networks

BEST FOR

  • Government and regulated organisations needing models in their own environment
  • Engineering teams improving search and retrieval quality
  • Enterprises wanting a private AI workspace without public cloud chat tools

TYPICAL USES

Improving internal search results with RerankBuilding retrieval systems over company documents with EmbedDeploying a private AI workspace for staff with NorthRunning language models inside a secure or disconnected networkClassifying and extracting data from business documents

002 / THE HONEST TAKE

STRENGTHS

  • +Private, on-premises and air-gapped deployment options
  • +Rerank and Embed models are strong building blocks for search
  • +North gives non-technical staff an agent workspace
  • +Free trial keys for development and testing

WATCH OUTS

  • –Smaller consumer presence and community than larger model makers
  • –Private deployment needs significant technical capability
  • –Enterprise products require sales engagement and custom pricing

003 / FOR AUSTRALIAN BUSINESSES

Can run onshore

Not by default, but it can be kept in Australia by self-hosting or using it through an Australian cloud region.

With private, on-premises or air-gapped deployment, data stays wherever you run Cohere's models, including Australian data centres. Cohere's hosted API does not advertise an Australian region, and cloud marketplace availability varies by model and region. Ask Cohere or your cloud provider which models can run in Australia and what logging applies to hosted API calls.

PRIVACY ACT & PRACTICAL NOTES

Cohere's appeal for Australian organisations is control. Because its models can run inside your own cloud account, on-premises or in an air-gapped network, government agencies, defence suppliers and regulated firms can keep data inside an Australian environment they manage. Its models are also offered through major cloud marketplaces, but regional availability varies, so check whether your chosen model runs in a Sydney or Melbourne region before relying on it. Private deployment needs in-house machine learning and infrastructure skills.

004 / WHAT IT COSTS

Freemium

Developers get free trial API keys with rate limits that cannot be used for production. Production API use is pay-as-you-go per token, while Model Vault offers dedicated instances billed hourly or monthly. North and private deployments are priced through enterprise agreements. See cohere.com/pricing.

Check current prices on cohere.com ↗︎

005 / CONNECTS WITH

Amazon BedrockAmazon SageMakerMicrosoft FoundryOracle Cloud InfrastructureLangChain

Want it wired into your CRM, inbox or Xero? That's what we build.

007 / QUESTIONS

Anything else is quicker to answer on a call. Thirty minutes, free, with the people who would build it.

Yes, if you deploy them privately. Cohere supports running its models in your own virtual private cloud, on-premises or in an air-gapped network, so an Australian data centre or cloud region works. Through cloud marketplaces, availability in Sydney or Melbourne depends on the specific model, so check before designing around it.

We set up and connect tools like this for Australian businesses, with the privacy settings done right from day one.

Book a free call→