Planning to use the Kimi API across a company or inside a product? There is no secret enterprise price list. You start from the same public per-token prices, and large customers can negotiate discounts, higher limits and a service agreement. This page explains how it works and how to estimate your bill.
Standard limits rise with your total top-ups. Beyond the top tier, limits are agreed by contract.
From $0.95 in and $4 out per million tokens, up to $3 and $15 for Kimi K3.
Volume discounts need enterprise verification and a signed contract.
Enterprise customers also get higher limits, support and an SLA.
Any company can use the standard Kimi API today by topping up a balance. An enterprise arrangement sits on top of that. Kimi's help centre lists four extras:
More requests at once and per minute than the standard tiers allow.
A technical support channel for your team.
Agreed service levels, written into your contract.
Priority access to new features.
Enterprise-verified customers can sign a standard or customized service contract.
After signing, you can top up by corporate bank transfer.
High-volume customers can apply. The level depends on committed usage and the length of the agreement.
From Kimi's help centre page on business cooperation. It gives no discount percentages, minimum spend or SLA figures.
These are the public prices every customer pays unless a contract says otherwise. Billing is pay as you go, with no subscription.
| Model | Input | Repeated input | Output | Context |
|---|---|---|---|---|
| kimi-k3 | $3.00 | $0.30 | $15.00 | 1,048,576 |
| kimi-k2.7-code | $0.95 | $0.19 | $4.00 | 262,144 |
| kimi-k2.7-code-highspeed | $1.90 | $0.38 | $8.00 | 262,144 |
| kimi-k2.6 | $0.95 | $0.16 | $4.00 | 262,144 |
On a small screen, swipe the table sideways to see every column.
Full detail: Kimi API pricing · Web Search API pricing
For a business, limits often matter more than price. Standard limits depend on how much your account has topped up in total.
| Tier | Total top-ups | At once | Requests a minute | Tokens a minute | Tokens a day |
|---|---|---|---|---|---|
| Tier 0 | $1 | 1 | 3 | 500,000 | 1,500,000 |
| Tier 1 | $10 | 15 | 100 | 2,000,000 | Unlimited |
| Tier 2 | $20 | 40 | 100 | 3,000,000 | Unlimited |
| Tier 3 | $100 | 50 | 200 | 3,000,000 | Unlimited |
| Tier 4 | $1,000 | 60 | 200 | 4,000,000 | Unlimited |
| Tier 5 | $3,000 | 100 | 300 | 5,000,000 | Unlimited |
| Enterprise | By contract | Higher limits, agreed with sales | |||
On a small screen, swipe the table sideways to see every column.
From Moonshot's rate-limit page, read on 8 October 2026.
Enter your expected traffic. The result uses list prices, so treat it as the figure before any discount.
Worked out as 30 days at list prices, before tax. It leaves out Kimi K3's charge for saving text to the cache, web search calls and any discount. Thinking counts as output, so allow extra for reasoning models. Average traffic is not peak traffic: plan your limits around your busiest minute.
Most savings come from how you use the API, before any negotiation.
Repeated input costs up to 90% less. Keep the fixed part of each prompt identical and at the start.
Send routine work to Kimi K2.6 and keep Kimi K3 for the hard cases. Output is almost four times cheaper.
Set a token limit on every request. Kimi K3's default cap is very high.
Use a lower reasoning effort on Kimi K3 for simple jobs.
Jobs that do not need an instant answer can go through the batch service at a lower price.
Once usage is predictable, ask sales about a volume discount.
Our earlier research for this site found batch jobs priced at 60% of standard for Kimi K2.7 Code and Kimi K2.6, with Kimi K3 not included. We did not re-check that figure today, so confirm it in Moonshot's docs.
Moonshot's own API is not the only route. Because Kimi's models have open weights, larger companies have three choices.
| Route | How you pay | Good for | Watch out for |
|---|---|---|---|
| Moonshot's API | Per token, or by contract | The newest models and features first | The service is run from China. Check your data rules |
| A cloud provider | Through your cloud bill | Companies that must keep data with their existing cloud. Kimi K3 is on Amazon Bedrock | Prices, limits and features are set by the provider and can differ |
| Run it yourself | Your own hardware and staff | Full control over data and uptime | Kimi K3 is very large. Moonshot suggests 64 or more accelerators |
On a small screen, swipe the table sideways to see every column.
Cloud and self-hosting details were collected earlier for this site and not re-checked today. More: Kimi K3 API · Kimi K3
The first four are what Kimi's help centre asks for. Use the estimator above for the usage figure.
Kimi's public pages do not answer these, so put them to sales and get the answers in writing.
What discount at what committed volume? What happens to unused commitment? Are invoices and tax documents provided? Can prices change during the term?
What uptime does the SLA promise, and what is the remedy if it is missed? What are our limits at peak? How fast does support respond?
Where is our data processed and stored? Is it used for training? How long is it kept? Is there a data processing agreement?
Kimi's help centre says its model has completed the required algorithm filing in mainland China, and that sales can supply compliance material. Have your own legal and security teams review the terms.