Kimi API Pricing and Moonshot Plan Guide
Kimi is offered through Moonshot’s China and global services as well as third-party channels. API token pricing and Kimi application subscriptions are separate products, so this guide presents API costs first and links to the relevant plan comparisons.
Current API price snapshot
Up to 12 tracked language models are shown. Channels are selected by lowest input rate normalized to USD, not lowest total cost. Examples use 1M input plus 250K output tokens, without additional cache, batch or context-tier adjustments or tool fees. Processing-mode variants are not interchangeable; no total is shown when output pricing is unknown.
| Model | Lowest-input-rate channel | Input / 1M | Output / 1M | Example cost (USD) | Channels | Selected price verified |
|---|---|---|---|---|---|---|
| Kimi K3 Batch | OpenRouter | $2.28 | $11.40 | $5.13 | 1 | Oct 17, 2026 |
| Kimi K3 | OpenRouter | $0.30 | $11.94 | $3.288 | 5 | Oct 17, 2026 |
| Kimi K2.7 Code | XycAi | ¥3.52 | ¥14.64 | $1.0411 | 4 | Oct 17, 2026 |
| Kimi | OpenRouter | $0.27 | $14.00 | $3.77 | 1 | Oct 17, 2026 |
| Kimi K2.6 | XycAi | ¥3.52 | ¥14.64 | $1.0411 | 4 | Oct 17, 2026 |
| Kimi K2.5 | OpenRouter | $0.50 | $2.50 | $1.125 | 1 | Oct 17, 2026 |
| Kimi K2 Thinking | OpenRouter | $0.60 | $2.50 | $1.225 | 1 | Oct 17, 2026 |
| Kimi K2 | OpenRouter | $0.57 | $2.30 | $1.145 | 1 | Oct 17, 2026 |
| Kimi K2.7 Code Highspeed | 月之暗面 (Moonshot/Kimi) | ¥13.00 | ¥54.00 | $3.8406 | 1 | Oct 17, 2026 |
| Kimi K2.6 Fast | Fireworks AI | $2.00 | $8.00 | $4.00 | 1 | Aug 28, 2026 |
| Kimi K2.7 Code Fast | Fireworks AI | $1.90 | $8.00 | $3.90 | 1 | Aug 28, 2026 |
| Kimi K3 Us | Fireworks AI | $4.50 | $22.50 | $10.125 | 1 | Oct 17, 2026 |
China and global services
The China and global services may differ in account registration, payment currency, network access, model rollout, and documentation. Treat them as separate channels even when the model family name matches.
API versus Kimi subscriptions
API billing is appropriate for software integration and scales with token use. Kimi subscriptions cover product features and usage allowances in the application; compare those on the linked plan page.
Cost formula: currency, cache and context length
Kimi cost = ordinary input ÷ 1M × I + cache-hit input ÷ 1M × C + output ÷ 1M × O + applicable cache-write or storage charges + billable web-search usage × its tool rate. Use the China site’s CNY rate card or the global site’s USD rate card. For models with context-length pricing, determine the tier from the full request, including history and documents; automatic context selection and named context-length variants can lead to different applicable rates. Use the exact model’s current rules rather than a single rate for every Moonshot or Kimi request.
Usage example: document questions with cached context
Assume 1,000 document-Q&A requests with measured totals of 3M input tokens, including 1M cache hits, and 0.5M output tokens. Assume every request stays within the selected model’s same context-price tier, uses a cache mechanism without separate write or storage charges, and makes 40 billable web-search calls. The illustration costs 2 × I + C + 0.5 × O + 40 × S in the chosen site’s currency. Search-result tokens are already included in the input total. If your model bills cache writes or storage, add those charges; if some requests enter another context tier, calculate that subset separately.
Easy-to-miss costs: search, context growth and recharge limits
Web search can add a tool fee and more input tokens when results are sent back to the model. Repeated document content is still billable input unless it qualifies for a cache hit, and longer history can change a context-priced model’s rate. Check cache lifetime and any write or storage line items. Recharge or account tiers determine rate limits under the selected site’s rules; prepaid balance does not guarantee unlimited concurrency. Compare China and global invoices in their original currencies before conversion.
How to use this data
- Recalculate with your real input/output token mix.
- Check region, rate limits, and payment method.
- Recheck official sources before purchasing or switching.
Frequently asked questions
Can I reuse China-site prices for the global API?
Use each site’s own model rates and billing currency: CNY for China and USD for global. Compare totals only after matching models and converting currencies.
Does automatic context selection mean one fixed price?
No. For context-priced models, inspect the applicable length tier or named variant. Other models may use a flat rate across their supported context; follow the exact model’s documentation.
Does context caching remove web-search charges?
No. Cache pricing concerns model input, while web search has its own tool fee. Search results passed into a later request also contribute to input usage.
Does recharging unlock unlimited requests?
No. Check the site’s recharge or account-tier rate limits and your current quota. API balance and Kimi application subscriptions are separate products.
Official sources and review date
Billing guidance reviewed: . Price verification dates are listed separately in the table.