Cohere
Enterprise AI: Private, Secure, Customizable
Last reviewed Mar 14, 2026
Cohere is an enterprise AI company founded in 2019 that provides private, secure, and customizable AI solutions for businesses. The company offers the Command family of language models, the Transcribe speech-to-text model, retrieval tools like Embed and Rerank, and the North enterprise AI platform, with flexible deployment options including cloud API, VPC, on-premises, and the Cohere-managed Model Vault.
LLM API Pricing
Pay-per-token pricing. Prices shown per 1M tokens.
Prices last updated: August 22, 2026
| Model | Input/1M | Output/1M |
|---|---|---|
| $0.037 | $0.150 | |
| $0.150 | $0.600 |
Cohere pricing by model
Input, output, and batch rates, plus alternatives, for one model at a time.
Pros & Cons
Advantages
- Strong enterprise security and compliance focus
- Flexible deployment options including on-premises
- Excellent multilingual support with Aya models
- Competitive pricing for high-quality models
- Purpose-built retrieval and reranking models
- Model customization on proprietary data
Limitations
- Smaller model ecosystem compared to OpenAI or Anthropic
- Less name recognition in consumer AI space
- Enterprise features may be overkill for smaller projects
- No image, video, or audio generation models
- Platform products (North, Compass) have no published pricing — enterprise sales contact required
Key Features
Command Model Family
High-performance generative models combining reasoning, tool orchestration, and multimodal intelligence, supporting 49 languages
North Mini Code
Agentic coding model built for practical software engineering tasks
Aya Multilingual Models
Research-grade multilingual models (8B and 32B) excelling across diverse languages
Transcribe
High-fidelity speech-to-text model supporting 14 languages with robust real-world conversation handling
Embed & Rerank
Multimodal semantic search and relevance optimization for retrieval-augmented generation
Flexible Deployment
Deploy via cloud API, virtual private cloud, on-premises, or Cohere-managed Model Vault
Enterprise Platform (North)
Enterprise AI platform for workplace productivity with customizable AI agents and multi-step workflow automation
Compass
Intelligent search and discovery system with pre-built data connectors, document parsing, and a managed index
Model Customization
Fine-tune models on proprietary data for domain-specific applications
Pricing Options
| Option | Details |
|---|---|
| Pay-as-you-go | Production API keys billed per million input and output tokens, invoiced monthly or once an outstanding balance threshold is reached |
| Trial API Key | Free rate-limited key created automatically with every account, for development and testing but not commercial use |
| Model Vault | Dedicated managed inference priced per instance by model and performance tier, billed hourly or via monthly and annual commitments, with Fixed or Flex plans |
| Enterprise | Custom pricing for the North and Compass platforms, model customization, and private or on-premises deployments |
Availability & Support
Regions
Global API access with deployment options including AWS, GCP, Azure, and on-premises installations
Support
Documentation, API reference, release notes, cookbooks, LLM University courses, Discord community, and enterprise support options
Getting Started
- 1
Create Cohere account
Sign up at dashboard.cohere.com
- 2
Get API key
Generate a trial or production API key from the dashboard
- 3
Install SDK
pip install cohere (Python) or npm install cohere-ai (TypeScript)
- 4
Make first API call
Call the Chat, Embed, or Rerank endpoint with your API key, or experiment in the dashboard Playground
Frequently Asked Questions
What GPU types does Cohere offer?
Check the pricing table above for Cohere's current GPU availability and pricing.
How do I get started with Cohere?
Create Cohere account, Get API key, Install SDK, Make first API call
What are Cohere's main advantages?
Cohere's main advantages include: Strong enterprise security and compliance focus, Flexible deployment options including on-premises, Excellent multilingual support with Aya models, Competitive pricing for high-quality models, Purpose-built retrieval and reranking models, Model customization on proprietary data.
What are Cohere's limitations?
Cohere's main limitations include: Smaller model ecosystem compared to OpenAI or Anthropic, Less name recognition in consumer AI space, Enterprise features may be overkill for smaller projects, No image, video, or audio generation models, Platform products (North, Compass) have no published pricing — enterprise sales contact required.
Compare Providers
Find the best prices for the same GPUs and models from other providers