kimi-k3

Moonshot·Usage-based billing

Multimodal Kimi model with 1M context and toggleable max-effort thinking for long-horizon agent work

Pricing

Supply channels

OfficialSales ratio 1x

Tiered price table

base

Input

$20 / 1M

Output

$100 / 1M

Cache Read

$2 / 1M

Cache Write

$20 / 1M

Cache Write (1h)

$40 / 1M

Tier	Input	Output	Cache Read	Cache Write	Cache Write (1h)
base	$20 / 1M	$100 / 1M	$2 / 1M	$20 / 1M	$40 / 1M

Hybrid cloudSales ratio 0.75x

Tiered price table

base

Input

$20 / 1M$15 / 1M

Output

$100 / 1M$75 / 1M

Cache Read

$2 / 1M$1.5 / 1M

Cache Write

$20 / 1M$15 / 1M

Cache Write (1h)

$40 / 1M$30 / 1M

Tier	Input	Output	Cache Read	Cache Write	Cache Write (1h)
base	$20 / 1M$15 / 1M	$100 / 1M$75 / 1M	$2 / 1M$1.5 / 1M	$20 / 1M$15 / 1M	$40 / 1M$30 / 1M

Special offerSales ratio 0.35x

Tiered price table

base

Input

$20 / 1M$7 / 1M

Output

$100 / 1M$35 / 1M

Cache Read

$2 / 1M$0.7 / 1M

Cache Write

$20 / 1M$7 / 1M

Cache Write (1h)

$40 / 1M$14 / 1M

Tier	Input	Output	Cache Read	Cache Write	Cache Write (1h)
base	$20 / 1M$7 / 1M	$100 / 1M$35 / 1M	$2 / 1M$0.7 / 1M	$20 / 1M$7 / 1M	$40 / 1M$14 / 1M

Model

ProviderMoonshot

TypeUsage-based billing

Endpoints

openaiopenai-response