AI Token Price
Moonshot AI

Moonshot AI · Kimi family · released 16 Jul 2026

Kimi K3 API pricing

Moonshot AI's Kimi models. Kimi K3, released 16 July 2026, costs $3 / $15 per 1M tokens with a 1M-token context window.

Kimi K3 prices

Input
$3.00
per 1M tokens
Output
$15.00
per 1M tokens
Cached input
$0.30
90% off input
Batch
—
no discount offered
Context window
1M
tokens
Verified 27 Sep 2026Kimi API pricing Compared with Kimi K2.6: 216% more on input, 275% more on output

What it costs in practice

Kimi K3 at list prices, no batch discount.

What Kimi K3 costs for three workloads
1,000 chat replies1,500 tokens in and 300 out each$9.00
Summarising 100 long reports20,000 tokens in and 800 out each$7.20
A month of a coding agent400 tasks, 85% of input cached$649

The coding-agent example assumes 85% of input is served from cache, typical for agents that resend the same context each step. Model your own workload

Details

API model IDkimi-k3
Released16 Jul 2026
WeightsOpen, 2800B total and 104B active parameters
Arena text score1488 (25 Sep)
StatusCurrent

Checked 27 Sep 2026 against Kimi API pricing and Morph (release date) .

Other Kimi versions

Still available through the API

Other Kimi models and prices per million tokens
Model Input / 1M Output / 1M Cached input Context Released
Kimi K2.6Previous
Open weights
$0.95 $4.00 $0.16 256K 20 Apr 2026

Kimi K2.6 details

Kimi K2.6

Previous

$0.95 / $4.00 per 1M tokenscached $0.16 · 256K context

Released 20 Apr 2026 · kimi-k2.6

Newer option: Kimi K3 at $3.00 / $15.00. Checked 27 Sep 2026 against Kimi API pricing .

Compare Kimi K3

Compare with any model

Price history

All changes
  • Moonshot AI
    Moonshot launched Kimi K3 at $3 / $15. Source

Moonshot AI pricing rules

  • Cached input is billed at a reduced rate.

All Moonshot AI models

Questions

How much does Kimi K3 cost?

As of 27 Sep 2026, Kimi K3 costs $3 per million input tokens and $15 per million output tokens, and $0.30 per million cached input tokens. Prices checked 27 Sep 2026.

What is the cheapest Kimi model?

Kimi K2.6, at $0.95 / $4 per million input / output tokens.

What is Kimi K3's context window?

1M tokens.

Does Kimi K3 support prompt caching?

Yes. Repeated input read from the cache costs $0.30 per million tokens, 90% less than normal input.