Kimi official global
Taxes may be added at checkout. The official page describes flat K3 token pricing rather than context-length tiers.
Kimi API pricing
Kimi API pricing
Compare K3 pricing without erasing the route behind it. Official global, official China and third-party providers keep their original currency, cache rules and evidence source.
Pricing snapshot — July 28, 2026
DECLAREDPrices below are provider declarations, not a Modroam guarantee. Live pricing, taxes, promotions, availability and account eligibility can change. Follow the source link before committing spend.
Route-by-route comparison
There is no universal Kimi sticker price.
Kimi does not necessarily have one universal sticker price. Official Global and China routes can use different billing systems, and third-party providers add another pricing layer.
| Route | Market | Currency | Cache read | Input | Output | Evidence | Source | Last checked |
|---|---|---|---|---|---|---|---|---|
| Kimi official globalOfficial managed APIkimi-k3 | Global platform | USD | $0.30 / 1M cached input tokens | $3.00 / 1M input tokens | $15.00 / 1M output tokens | DECLAREDofficial-declared | Kimi global K3 pricing ↗ | July 28, 2026 |
| Kimi official ChinaOfficial managed APIkimi-k3 | China platform | CNY | ¥2.00 / 1M cached input tokens | ¥20.00 / 1M input tokens | ¥100.00 / 1M output tokens | DECLAREDofficial-declared | Kimi China K3 pricing ↗ | July 28, 2026 |
| Fireworks AIThird-party inference provideraccounts/fireworks/models/kimi-k3 | Provider-specific | USD | $0.30 / 1M cached input tokens | $3.00 / 1M input tokens | $15.00 / 1M output tokens | DECLAREDprovider-declared | Fireworks K3 model page ↗ | July 28, 2026 |
Scroll horizontally on narrow screens. Prices remain in each source's original currency; no exchange-rate conversion is applied.
Kimi official China
These are original CNY list prices shown by the China platform. They are not a converted copy of the global USD price.
Fireworks AI
Fireworks lists a serverless route and on-demand deployment. Provider features and data handling remain separate from Moonshot's official API.
DECLAREDPublished by the named provider.OBSERVEDIndependently captured usage or invoice evidence. No observed price is claimed in this table.
Read the bill correctly
Four checks before comparing totals
Cache hit or cache miss?
Repeated prompt context may qualify for a lower cached-input rate. Confirm the provider's cache rules and whether your workload actually hits them.
Which billing currency?
Keep USD and CNY prices in their original official currency. Exchange-rate conversions are estimates, not the provider's invoice price.
Which route and region?
An account on the global platform, China platform or a third-party provider can have different endpoints, terms, limits and payment methods.
What else is billed?
Token charges may not cover taxes, web-search calls, dedicated capacity, egress or enterprise commitments. Check the live source before purchase.
Coding Plan or API?
Choose by workload shape, not sticker price alone
A subscription can fit frequent interactive coding while API billing can fit programmatic execution. Keep both routes visible when your work spans the two.
Interactive coding
Kimi Coding Plan candidate
Programmatic, backend or batch work
Kimi API candidate
Mixed interactive and automated work
Coding Plan plus API backup
No real usage data
Record usage; do not calculate break-even yet
K2 and K2.5 queries
Do not apply an old generation's price to K3.
Kimi K2, K2.5, K2.6, K2.7 Code and K3 are separate model entries. Historical pricing can help explain an invoice, but the current platform model table should decide a new purchase.
This stable pricing URL will track the current model family instead of creating isolated pages for every release and price wording variant.
Access request
Need Kimi access or API help?
Tell us whether you need platform access, an API route, a coding plan or setup support. We’ll check availability, scope and a test path.
Same model ≠ same route ≠ same commercial terms.
Confirm the account, endpoint, currency, cache rule, provider terms and live source before committing spend.
Kimi payment decision FAQ
Monthly, annual, or API?
Buy Kimi for what it can do for your workflow today—not for the model you hope it becomes next year.
Kimi's annual billing can reduce the effective price, but the discount also locks you into one provider, one quota system and a product that may continue to change.
Use a monthly plan first unless you already know that Kimi fits your real workload.
Quick decision table
Situation
You are new to Kimi or have not completed a real project with it
Recommendation
Start with one month
Situation
Your usage is irregular or you only need occasional high-volume calls
Recommendation
Pay-as-you-go API
Situation
You have used the same plan consistently and expect to use it for at least 10 months
Recommendation
Consider annual billing
Situation
You already hit the limits of your current tier
Recommendation
Compare an upgrade, Extra Usage or a mixed-model workflow
Situation
You need multiple models or provider redundancy
Recommendation
API or multi-provider setup
Situation
Kimi is already part of a stable personal coding workflow
Recommendation
A Kimi Code membership may fit better
These are decision defaults, not availability or savings guarantees.
Annual commitment check
Four questions before choosing annual billing
- 1Have I completed a real project using the same monthly tier?
- 2Does its current quota cover a normal working week?
- 3Would I still use Kimi if there were no major model upgrade during the next six months?
- 4Do I expect to use it during at least 10 of the next 12 months?
If most of your answers are not clearly yes, stay monthly or use the API first.
Seven short answers before you pay
Is the Kimi annual plan worth it?
It can be, but only after the monthly plan has survived a real workload.
Use one full billing cycle to test:
- model quality on your own repositories;
- speed during your normal working hours;
- weekly and five-hour limits;
- how often you need another model for planning, frontend work or review.
The important decision is not whether Kimi may improve next year. It is whether the current product, quota and workflow already justify locking in for a year.
How much do I actually save with annual billing?
At the currently published prices, annual billing costs approximately 9.5–9.6 months of monthly billing, producing a discount of roughly 20%.
| Plan | Monthly price | Annual total | 12 monthly payments | Approx. saving | Equivalent monthly payments |
|---|---|---|---|---|---|
| Moderato | $19/mo | $180 | $228 | $48 | 9.5 months |
| Allegretto | $39/mo | $372 | $468 | $96 | 9.5 months |
| Allegro | $99/mo | $948 | $1,188 | $240 | 9.6 months |
| Vivace | $199/mo | $1,908 | $2,388 | $480 | 9.6 months |
The saving is real, but it should be weighed against model, quota and provider lock-in. Prices and plan structures can change, so verify the checkout page before purchasing.
Should I choose a Kimi subscription or the pay-as-you-go API?
Choose a Kimi Code membership when you have a recurring personal coding workflow and the included quota matches your normal usage.
Choose the Kimi Open Platform API when you need:
- product or application integration;
- irregular usage;
- usage-based billing;
- multiple model or provider routes;
- more control over your own application architecture.
Kimi Code and the Open Platform are separate systems with different keys and endpoints; their credentials are not interchangeable.
Review Kimi API access and setup →How do Kimi credits and usage limits reset?
Kimi membership features use a shared credit pool that resets with the monthly subscription cycle. Using credits in one membership feature can reduce what remains for other features.
Kimi Code also has its own weekly allowance and rolling five-hour rate limit. Its Benefits page describes the coding allowance as refreshing every seven days, and unused weekly Kimi Code quota does not roll over into the next cycle.
This means “credits remaining” and “temporarily rate-limited” are not always the same problem.
What happens when I run out of credits?
Depending on which limit you reach, you can:
- wait for the relevant limit to refresh;
- move to a higher tier;
- enable Extra Usage if your account is eligible;
- route bounded tasks to another model or API.
Kimi describes Extra Usage as actual-consumption billing priced close to the official Open Platform API, not as a permanently identical rate. Compare the live Extra Usage terms, official API price and your real usage.
Do not assume a subscription is always cheaper. Compare your actual Extra Usage and API spending.
Can I use my Kimi plan with VS Code, Claude Code, OpenCode or Codex?
Kimi currently lists Kimi Code CLI, Claude Code and Roo Code as supported Kimi Code tools. Roo Code is the listed VS Code extension.
OpenCode, Codex and other community tools are not on the current supported-tool list. They may work through custom providers or compatible endpoints, but they should not be described as officially supported unless Kimi adds them to its documentation.
For unsupported or community-tested tools, use a separate setup guide rather than treating compatibility as guaranteed.
Can I cancel, upgrade or downgrade an annual plan?
Cancelling stops future auto-renewal; it does not automatically refund the unused portion of an annual plan. Benefits remain active through the current paid period.
Upgrades may take effect immediately. Kimi says it calculates the unused portion of the current plan on a prorated basis as part of the upgrade. Downgrades cannot be applied mid-cycle; the documented path is to cancel the current subscription, wait for it to expire and then choose a lower tier.
Activated paid services may be non-refundable and non-transferable unless applicable law requires otherwise. Check the current account and payment pages before making an annual commitment.
Plan structure notice
Kimi says membership benefits and Kimi Code benefits are expected to become separately purchasable under a future membership system. Check the current subscription page before making an annual commitment.
Evidence boundary: official documentation supplies current product and policy facts; annual savings are Modroam calculations from the published prices; community discussion informs the questions only and is not used as a plan-rule source.