Captured October 26, 2026 at 1440px wide.
Cohere plans and prices
| Plan | Price |
|---|---|
| NorthHighlightedOur all-in-one AI platform built to empower humans, automate tasks, and innovate safely across your enterprise. | Get in touch for custom enterprise pricing. |
| CompassOur intelligent search and discovery system designed to surface insights from across your business. | Get in touch for custom enterprise pricing. |
As printed on the page captured October 26, 2026.
How does Cohere pricing work?
Each product line carries its own value metric. North, 'Our all-in-one AI platform', and Compass, an 'intelligent search and discovery system', both read 'Get in touch for custom enterprise pricing.', so the platform is sold as a contract. Model Vault is priced by reserved capacity: 'Pricing is determined per instance based on the selected model and its performance tier.' Embed 5 Fast and Embed 5 Pro both run $3.00 hourly and $2,000 monthly on Small and $5.00 and $3,250 on Medium, Rerank 5 Fast, Rerank 5 Pro and Parse 5 Medium run $4.00 and $2,500, and Parse 5 XL runs $7.00 and $4,300. The API runs on tokens and search units, with 'Production API key will be charged on a pay-as-you-go basis.' Three metrics fit three ways an enterprise adopts AI: as a platform, as dedicated infrastructure, or as an API.
The motion is sales-led at the front with a self-serve developer path behind it. A banner opens with 'Rerank 5 is here: sharper search, fewer tokens.', and the hero, 'Empower your team to put AI to work', offers 'Contact sales' and 'Explore private deployments'. A toggle splits the page into Workplace systems, Generative models, Advanced retrieval models and Coding models. North and Compass each list four scope bullets and 'Request a demo'; Model Vault offers 'Cohere Dashboard' beside 'Speak to sales'. Twelve FAQ questions cover trial and production keys, organizations, billing, tokens, legacy rates and Rerank search units, followed by a Fujitsu testimonial and a demo request form that asks for deployment preference across AWS, Azure, OCI, GCP and private deployment.
Under the hood, the FAQ maps a clear land-and-expand path. Every new account automatically receives a Trial API key, and trial keys serve free, rate-limited calls for non-production use. Production keys are pay-as-you-go, and 'Your bill will be issued at the end of every calendar month or when you reach $250 in outstanding balances', a threshold invoice that caps credit exposure on new accounts. Model Vault then turns usage into reserved capacity, billed 'hourly or through longer-term commitments (monthly or annual)' with 'Fixed or Flex pricing plans available'. The Fast and Pro variants of Embed 5 and Rerank 5 print identical rates per tier, which points to an instance price set by the compute behind each performance tier. Rerank's unit is defined to the token: 'one query with up to 100 documents', with documents over 500 tokens split into chunks that each count. Free trial, metered production, then committed capacity is a ladder worth borrowing for any model provider.
What does Cohere's pricing page do well?
- A four-way toggle splits pricing by buying job: Workplace systems, Generative models, Advanced retrieval models and Coding models
- The Model Vault table prints both the hourly and the monthly rate per instance, so a buyer can price a test and a steady deployment from one row
- Fast and Pro model variants share one rate per performance tier, so the table reads as a short ladder from $3.00 to $7.00 an hour
- The paragraph under the table states the billing mechanics plainly, including hourly billing or 'longer-term commitments (monthly or annual)'
- The FAQ defines a Rerank search unit exactly: 'one query with up to 100 documents to be ranked', with longer documents split into 500-token chunks
What does Cohere's pricing page reveal about its pricing strategy?
- Three value metrics match three ways to adopt AI: platform contracts, reserved instances and pay-as-you-go tokens
- Invoicing at month end or at $250 of outstanding balance caps credit exposure on new pay-as-you-go accounts
- Model Vault converts usage into reserved capacity with hourly, monthly or annual terms, the step from API spend to committed revenue
- Embed 5 Fast and Pro print the same $3.00 and $5.00 rates, so the instance price tracks the performance tier and the model variant stays a product choice
- Deployment preference on the demo form, from AWS to Private Deployment, signals that where the model runs is part of every enterprise price
Learn with Orb
- GuideAn expert's guide to enterprise pricing for SaaS North and Compass are sold on custom enterprise contracts behind Request a demo, and Orb's enterprise pricing guide covers the main enterprise models and what to weigh when prices are set deal by deal.
- DocsConfiguring plans: minimums and plan phases Model Vault bills reserved instances hourly or on monthly and annual commitments; Orb represents a dollar commitment as a minimum and uses plan phases to step committed spend up over time.
- GuidePricing AI agents: plans, costs and monetization models The API meters pay-as-you-go tokens and Rerank search units defined down to 100 documents per query, the per-call and per-token metering that Orb's guide to pricing AI walks through.
Questions about Cohere pricing
How much does Cohere cost?
Cohere's pricing starts at $3.00/instance/hr. The page shows 2 plans, including a free plan, with the top plan priced through sales.
What pricing model does Cohere use?
Cohere prices by usage, combined with quoted plans. The page is built around talking to the sales team. Its plans are aimed mainly at enterprises.
Does Cohere have a free plan?
Yes, Cohere has a free plan. Paid plans start at $3.00/instance/hr.
Does Cohere offer an annual discount?
Cohere offers monthly and annual billing; the page doesn't state the discount. The listed prices start at $3.00/instance/hr. Terms for the plans priced through sales are set with the sales team.
How is Cohere's pricing page laid out?
The page shows 2 plans, highlights North, and includes an FAQ. The breakdown above marks 14 points on it: 6 things it does well and 8 insights into the pricing.
How is a search unit defined for Cohere Rerank pricing?
Cohere defines a single search unit as "one query with up to 100 documents to be ranked." Any document over 500 tokens, including the length of the search query, is automatically split into multiple chunks. Each chunk counts as an individual document toward the total ranked for that search, which keeps pricing consistent for longer documents.
When does Cohere bill for Production API key usage?
Production API key calls are billed on a pay-as-you-go basis. The page states, "Your bill will be issued at the end of every calendar month or when you reach $250 in outstanding balances." Calls made from a Trial API key are free, with rate limits for non-production use.
What is Cohere's pricing for legacy Command models?
For existing customers, Command is $1.00/1M tokens for input and $2.00/1M tokens for output, and Command-light is $0.30/1M tokens for input and $0.60/1M tokens for output. Command R 03-2024 runs $0.50/1M input and $1.50/1M output tokens. Command R+ 04-2024 is $3.00/1M input and $15.00/1M output tokens, while Command R+ 08-2024 is $2.50/1M input and $10.00/1M output tokens.
From the page as captured October 26, 2026. Current prices are on cohere.com
AssemblyAI
$0.15/hrAssemblyAI bills speech-to-text by the hour of audio, the unit a transcription team already uses, and the default view shows two rates: Universal-3.5 Pro at $0.21 /hr and Universal-2 at $0.15 /hr. A selector across eight APIs, grouped under TRANSCRIBE, UNDERSTAND and INTERACT, swaps the rate table, so each product is priced in its own unit, from audio hours to LLM tokens to input characters. The FAQ carries the billing mechanics in plain words, including the new Text-to-Speech API at $15 per 1M characters.
Baseten
From $0.01/1M tokensBaseten layers a three-tier service ladder over published rate sheets. Basic is '$0 per month, pay as you go', and Pro and Enterprise carry 'Volume discounts available' with 'GET A QUOTE'. Underneath sit three meters: Model APIs per 1M tokens across 19 model rows with Input, Cache Input and Output columns, opening with joey-1 at $0.01 input, dedicated GPU and CPU instances billed 'down to the minute', and the same GPU grid for training.
Deepgram
$0.0077/minDeepgram's page opens on "Free to start. Priced to scale." and three cards: Pay as you go with a Free $200 credit, Growth at $4,000+ / year in pre-paid credits for up to 20% off, and Enterprise at Custom pricing. A tabbed "Rates by product" section names the meter for each product, per second of audio, per 1,000 characters, or per minute of connection time, and a sixteen-question FAQ carries the billing mechanics, from hourly rates to the 10% overage premium.
Exa
$4/1k requestsExa prices its search API per thousand requests, with a rate for each endpoint: Lite Search $4, Auto Search $7, Deep Search $12, Contents $1 per 1k pages, Monitors $2 and Answer $5, plus $1 per 1k for each result above ten and for AI page summaries. Agent runs offer metered compute at $0.10 per ACU and six effort levels: five fixed from $0.012 to $1.00 per request and an Ultra level billed on usage. The plan cards divide access, throughput and compliance, and the rate card below them carries the price.
fal
From $0.0002/secondfal prices two products on one page with rate sheets in place of plans: GPU hours for serverless compute across six SKUs, and model APIs billed per output unit across video, image, audio and 3D. The "Units per $1" column turns seconds, tokens, images, megapixels, characters, generations and credits into one number a buyer compares directly.
Fireworks AI
From $0.008/1M tokensFireworks prices three product lines on three meters: serverless inference per token, training per 1M training tokens, and on-demand deployments per GPU second. Serverless usage now draws down a pre-paid credit balance with Auto Reload and a monthly spend limit, and the training side carries a fine-tuning matrix that doubles cleanly at each step plus an eight-model Serverless Training API that meters prefill, cached prefill, sample and train tokens separately.