Kimi K3.1 Pro API pricing: official $2 / $10 per MTok
As of 2026-10-26, Moonshot's official pricing page lists kimi-k3.1-pro at $2.00 per million input tokens, $10.00 per million output tokens, and $0.20 per million cached input tokens; Highspeed is $4.00 input, $20.00 output, and $0.20 cached. The official Kimi Code updates page dates the series launch to 2026-10-23, and the official blog says the full weights will be released by 2026-11-04.
Updated 2026-10-26
Prices change: on QCode, Claude, GPT and Chinese models are billed per token — live rates on /models.
- Billed per token — live rates on /models
- Pay by card (Visa / Mastercard / AMEX), Apple Pay, Google Pay or crypto
- Self-service: top up and activate instantly
Four key points
Standard official price
The official USD pricing page: $2.00 input, $10.00 output, and $0.20 cached input per million tokens, before taxes. Page check date: 2026-10-24.
Official CNY page price
The official CNY page: ¥13.00 input, ¥68.00 output, and ¥1.30 cached input per 1M tokens, plus applicable taxes; it is priced separately from the USD page and not converted. Page check date: 2026-10-24.
Thinking mode only
The Kimi API docs: K3.1 Pro always has thinking enabled; the Kimi Code docs say disabling thinking returns a 400 error.
Weight deadline
The official blog: the full model weights will be released by 2026-11-04.
What K3.1 Pro is and when it launched
Moonshot's official blog calls Kimi K3.1 Pro the first model in the K3.1 series and a 3T-class open-weight model with a 1M-token context window; K3.1 Pro and K3.1 Pro Highspeed are available on Kimi on the web, Kimi Work, Kimi Code, and the Kimi API; the official Kimi Code updates page dates the K3.1 Pro series launch to 2026-10-23. The API model IDs are kimi-k3.1-pro and kimi-k3.1-pro-highspeed.
What the official pages say about price and status
The USD pricing page separates the standard and Highspeed models: the standard model is $2.00 input, $10.00 output, and $0.20 cached input; Highspeed is $4.00 input, $20.00 output, and $0.20 cached input, all per million tokens (MTok) before taxes. The CNY page gives a separate set of CNY prices: the standard model ¥13.00 / ¥68.00 / ¥1.30, and Highspeed ¥26.00 / ¥136.00 / ¥1.30. The USD and CNY pages are priced separately, with no conversion. Both pages state a rates-checked date of 2026-10-24.
Timeline
The official Kimi Code updates page dates the K3.1 Pro series launch to 2026-10-23; the official blog says K3.1 Pro and Highspeed are available on Kimi on the web, Kimi Work, Kimi Code, and the Kimi API.
The official USD and CNY pricing pages both give 2026-10-24 as the date their rates were checked; the USD page lists the standard and Highspeed USD prices.
The official blog (captured 2026-10-26) says the full model weights will be released by 2026-11-04.
Confirmed vs not published
Confirmed (verbatim in the official pages)
Launched 2026-10-23 (official Kimi Code updates page); model IDs kimi-k3.1-pro and kimi-k3.1-pro-highspeed; USD and CNY prices for the standard and Highspeed models; thinking mode only; weights by 2026-11-04.
Unverified and not published
The official pages give no exact weight release date beyond the by 2026-11-04 statement.
Standard vs Highspeed
kimi-k3.1-pro
Official USD price: $2.00 input, $10.00 output, and $0.20 cached input per million tokens.
kimi-k3.1-pro-highspeed
Official USD price: $4.00 input, $20.00 output, and $0.20 cached input; the official page says it runs about 2–3× faster than the standard model.
Calling and checking the price
① Pick kimi-k3.1-pro or kimi-k3.1-pro-highspeed on the Kimi API platform (the official blog's API note). ② Do not disable thinking: the official docs say K3.1 Pro always thinks, and the Kimi Code docs say disabling it returns a 400 error; reasoning_effort supports low, high, and max. ③ Check input, output, and cached-input charges on the official pricing page; the USD and CNY pages are priced separately.
Calling Kimi on QCode
As of 2026-10-26, QCode's /models lists kimi-k3 and does not list kimi-k3.1-pro. To call Kimi on QCode, use kimi-k3 from /models: the QCode docs say Chinese model families can use both the Anthropic API and the OpenAI Chat API, with a cr_ key from the dashboard. Billing is per token, with per-model rates on /models.
Frequently asked questions
What does Kimi K3.1 Pro cost through the API?
The official USD page: kimi-k3.1-pro costs $2.00 per million input tokens, $10.00 per million output tokens, and $0.20 per million cached input tokens; Highspeed costs $4.00 input, $20.00 output, and $0.20 cached, before taxes. The CNY page: the standard model ¥13.00 / ¥68.00 / ¥1.30, and Highspeed ¥26.00 / ¥136.00 / ¥1.30. Page check date: 2026-10-24.
When did Kimi K3.1 Pro launch?
The official Kimi Code updates page says 2026-10-23; the official blog says K3.1 Pro and Highspeed are available on Kimi on the web, Kimi Work, Kimi Code, and the Kimi API.
Can I turn thinking off to save money?
No. The Kimi API docs say K3.1 Pro always has thinking enabled, and the Kimi Code docs say disabling it returns a 400 error. You can adjust reasoning effort with reasoning_effort at low, high, or max.
When will the weights be released?
The official blog says the full model weights will be released by 2026-11-04; it gives no more specific date.
Is kimi-k3.1-pro available on QCode?
Not at present. The Kimi model on QCode's /models is kimi-k3; billing is per token, with per-model rates on /models.
What benchmark scores does Moonshot report?
Moonshot says K3.1 Pro scores 65.7% on Terminal-Bench 4.0, ahead of GPT-6 Astra at 58.2% and just behind Claude Opus 5.5 at 66.4%; these are Moonshot-reported results.
Sources
The official Moonshot blog and K3.1 Pro pricing pages (USD and CNY, captured 2026-10-26); the Kimi platform pricing page, model list, and K3.1 Pro quickstart (captured 2026-10-26); the official Kimi Code updates page (captured 2026-10-26); the QCode docs and /models page (captured 2026-10-26).
Try Kimi K3 on a pay-as-you-go pack for a few days
Not ready for a monthly plan? Pay-as-you-go Packs start at $4 for a one-time quota you use within 7 days; if you already have a Personal Plan, add a booster pack when you need more quota.
Related reading
Claude Code Pay-as-you-go Packs: No Subscription, 7 Days
Buy quota once and use it within 7 days, no subscription; extra packs add to the same key.
Kimi K3 status 2026: official price $3/$0.30/$15
Kimi K3's official price, context, and status on QCode.
Kimi K3 Status (Oct 2026): Live Since July, No K4 Listed
When K3 launched, its official price, and the fact that K4 is not yet on the official list.
Crypto payment guide
Nine coins, 56 networks, zero platform fee for AI coding API credits.
Moonshot details on this page were checked on 2026-10-26 against the official blog, pricing pages, and Kimi platform docs; QCode setup and model availability follow docs.qcode.cc and /models. Official changes are not announced separately.