Kimi K3.1 Pro: Access, API Pricing and the Weights Release

Kimi K3.1 Pro is available through Kimi's apps, coding tools and API. Choose the access route first: API token charges are separate from a product subscription, and the announced downloadable weights are still scheduled for November 4, 2026.

Official access and docs

Checked October 25, 2026 · Official-source review; no hands-on benchmark

Choose hosted access today

The official launch describes a 3T-class model with a one-million-token context window and a Highspeed variant. It lists web, mobile, Kimi Work, Kimi Code and API access. Full weights are promised by November 4; a planned release is not a download available at this check.

Use the app for an interactive task, Kimi Code for repository work, or the API when your own application needs to send requests. Benchmark and speed claims in the announcement are vendor results; this page does not rank the model from a hands-on test.

API costs: distinguish input, output and cache

The international API platform lists these US-dollar rates per one million tokens, checked October 25, 2026. Regional billing pages can use different currencies.

Scroll the table horizontally on a small screen to see all rates.

International Kimi API list rates, USD / 1M tokens
Variant	Input	Output	Cache write	Cache hit
K3.1 Pro	$2.00	$10.00	$2.00	$0.20
K3.1 Pro Highspeed	$4.00	$20.00	$4.00	$0.20

Worked example: 100,000 uncached input tokens and 5,000 billed output tokens cost (100,000 × $2 + 5,000 × $10) / 1,000,000 = $0.25 on the standard variant. Highspeed is $0.50 for the same token counts. This calculation excludes tools, taxes and other charges; it is not a measured bill.

Inspect actual billed output and reasoning usage. A short visible answer need not imply a small total token bill. Confirm the account's quota, currency and current tariff before submitting a large context.

Use the model ID for your access route

The launch uses API IDs kimi-k3.1-pro and kimi-k3.1-pro-highspeed. Kimi Code documentation instead lists k3.1-pro and k3.1-pro-highspeed for that service. Do not assume the identifiers are interchangeable.

The Code docs require thinking mode for K3.1 Pro and advise starting a new session when switching models. For a first evaluation, use a small repository task with known acceptance checks, inspect the returned diff, and compare usage before trying a long-running agent task. We have not called the API or tested a subscription.

What to verify before choosing

For long documents, include a question whose answer is near the end and check the cited passage. For code, verify the result with your project's existing checks. Compare standard and Highspeed using the same task and billed-token report; advertised speed alone does not establish lower cost per successful task.

Start from the official API platform or the product access links. Recheck the weights release separately if local deployment is a requirement.