We may earn a commission if you buy through links on this page. It never changes a score or a ranking. How we make money
There's a free plan; paid tiers are quoted.
For a team of one to five, Cohere’s Trial API key costs nothing to use, but it is rate-limited and not allowed for production or commercial work. Production API access is pay-as-you-go, with current model-specific rates not published in these pricing details. North and Compass require custom enterprise pricing. Model Vault lists hourly GPU rates, so your bill depends on how long you run it—not how many teammates have access.
01Every plan, line by line
| Plan | Billed annually | Billed monthly | Built for |
|---|---|---|---|
| Trial API key | — | Quote | Prototyping with Cohere APIs |
| Production API key | — | Quote | Production API workloads |
| North | — | Quote | Enterprise AI teams |
| Compass | — | Quote | Business search and discovery |
| Model Vault — L4 24GB | — | Quote | Managed model inference |
| Model Vault — A100 40GB | — | Quote | Managed model inference |
| Model Vault — A100 80GB | — | Quote | Managed model inference |
| Model Vault — MI300X 192GB | — | Quote | Managed model inference |
| Model Vault — H100 80GB | — | Quote | Managed model inference |
| Model Vault — H200 141GB | — | Quote | Managed model inference |
| Model Vault — H100 80GB encrypted | — | Quote | Confidential-compute inference |
| Model Vault — B200 192GB | — | Quote | Managed model inference |
| Model Vault — B200 192GB encrypted | — | Quote | Confidential-compute inference |
| Model Vault — GB200 192GB | — | Quote | Managed model inference |
Add-ons and overages
- Model Vault reserved capacity is available; contact sales for pricing.
- Further promotional pricing is available for monthly Model Vault commits.
- Legacy API rates for existing customers: Command $1.00/1M input tokens and $2.00/1M output; Command-light $0.30/$0.60; Command R 03-2024 $0.50/$1.50; Command R+ 04-2024 $3.00/$15.00; Command R+ 08-2024 $2.50/$10.00.
- Aya Expanse 8B and 32B API rates: $0.50/1M input tokens and $1.50/1M output tokens.
02What the plan grid doesn't tell you
Production API usage
The free Trial API key can’t be used commercially. Production access is pay-as-you-go, and current model-specific rates aren’t published in these pricing details.
Custom enterprise quotes
North and Compass don’t have public monthly or annual prices. Ask Cohere for a quote that reflects your intended use.
GPU hours add up
Model Vault bills per GPU per hour. Your total depends on runtime; reserved capacity pricing requires contacting sales.
Commit pricing
Further promotional pricing is available for monthly Model Vault commits, but Cohere doesn’t list those rates here.
03Which plan should you buy?
| If you are… | Get | Why |
|---|---|---|
| Solo prototype | Trial API key | Free API calls are useful for prototyping, but the key is rate-limited and not for production or commercial use. |
| Team of 3-5 in production | Production API key | This is the production API option; it is pay-as-you-go, and current model-specific rates aren’t published in these pricing details. |
| Team needing managed inference | Model Vault — L4 24GB | The listed on-demand rate is $3.00 per GPU per hour, or $1.00 through March 31, 2027 on promotion. |
04Is Cohere worth the money?
Cohere is worth considering if your small team wants to prototype AI or needs business search, agents, or managed inference. The free Trial API key keeps initial experimentation inexpensive, but it is not a production plan. For commercial use, plan around usage-based API billing or request a custom quote for North or Compass. Model Vault’s published hourly rates give you a starting point, but estimate GPU runtime before committing. If you only need a basic chat tool, Cohere’s enterprise focus may be more than you need.
Cohere pricing FAQ
Is Cohere free to use?
Cohere’s Trial API key allows free API calls, but it is rate-limited and not permitted for production or commercial use. Production API access is pay-as-you-go. North and Compass have custom enterprise pricing.
What is the cheapest Cohere production option?
The Production API key is the listed route for production API workloads, billed pay-as-you-go. Cohere doesn’t publish current model-specific rates in these pricing details, so ask for a quote or rate information based on your model and usage.
Does Cohere charge per seat?
Cohere lists API pricing as pay-as-you-go and Model Vault by GPU hour, rather than by seat. North and Compass use custom enterprise pricing. Cohere doesn’t publish a per-user price for those enterprise products.
How much does Model Vault cost?
Model Vault rates vary by GPU and are charged per GPU per hour. The listed on-demand rate for Model Vault — L4 24GB is $3.00 per GPU hour, with a promotional $1.00 rate through March 31, 2027. Your total depends on runtime.
