Is the Kimi API free?
Some eligible users get trial credit, but Kimi does not document it as unlimited free API access. Kimi’s English Help Center says newly registered users with a mainland China mobile number receive an RMB 15 API voucher, valid for three months after it is issued. The voucher is applied automatically and deducted before top-up balance, so eligible users can use it before adding paid balance for API models the voucher covers. A separate official Chinese-language trial page says the voucher cannot be used for Kimi K3; don’t count it toward a K3 budget. (kimi.com)
After a voucher is used up, API calls are charged against top-up balance. Kimi says requests return a 403 error when the balance is zero; its balance documentation likewise says inference API calls are unavailable when available balance is at or below zero. (kimi.com)
Kimi chat or membership access is separate from API billing. Free chat access does not, by itself, mean API calls are included. Third-party gateways may make their own offers, but those are not official Kimi API credits.
What can you budget for Kimi API use?
The Kimi K3 pricing entry lists these rates per 1 million tokens:
| Token category | Listed price |
|---|---|
| Cached input | $0.30 |
| Uncached input | $3.00 |
| Output | $15.00 |
These are the listed K3 rates, not a price for every Kimi model or third-party gateway. The available pricing material does not establish whether these rates apply to every account market or platform; confirm the rate shown for your account and selected model before budgeting. (kimi.ai)
Example: estimating a Kimi K3 request bill
For 1 million input tokens and 100,000 output tokens, using the listed K3 rates:
- All input uncached: (1 × $3.00) + (0.1 × $15.00) = $4.50.
- All input cached: (1 × $0.30) + (0.1 × $15.00) = $1.80.
This is an arithmetic illustration, not a guaranteed bill. It assumes the listed rates apply to your account and, for the second calculation, that all input qualifies for the cached-input rate. Actual token use, cache hits, and the rates applicable to your account affect the total.
For your own estimate, multiply representative input and output token counts by the selected model’s corresponding rates, then scale by expected request volume. Include system prompts and conversation history in the input estimate; don’t assume repeated text will always receive cached pricing.
Does a free or low-cost API make sense for production?
Free access and production readiness are separate questions. Check your account’s current limits, then test latency, error handling, output quality, and spend controls against your workload before relying on the API in production. The pricing and voucher information above does not establish a service commitment or prove production suitability. (platform.kimi.ai)
Bottom line
Kimi documents an RMB 15 voucher, valid for three months, for newly registered users with a mainland China mobile number. It is applied before top-up balance, but the official Chinese-language trial page says it cannot be used for Kimi K3. After the voucher is used up, calls require available top-up balance; at zero balance, Kimi says API requests return 403. The listed K3 rates are $3 per million uncached input tokens, $0.30 per million cached input tokens, and $15 per million output tokens. Confirm those rates in your account before calculating spend, and keep API charges distinct from chat access and third-party gateway offers. (kimi.com)
Sources
Kimi API Free Trial Credit Guide
Kimi API Free Trial Credit Guide (Chinese)
Check Balance — Kimi API Platform
Kimi K3 Pricing: Plans, Membership & API Costs