Kimi API pricing explained: K3.1 Pro, K3, K2.6

Kimi API pricing explained

Updated 2026-10-29. Official Moonshot prices, LMArena ranks.

Quick answer

Kimi K3.1 Pro is the model to buy: $2 input, $0.20 cached and $10 output per 1M tokens with a flat price across its 1M-token window, per Moonshot's official pricing page. It costs 33% less than the previous flagship Kimi K3 and ranks #6 on Code Arena against #13 for K3. The Highspeed version costs exactly double ($4 / $20) for about 2 to 3 times the speed. The K2 series is closed to new sign-ups from Oct 31, 2026.

Model	Input	Cached	Output	Context	Code Arena
Kimi K3.1 Pro Highspeed	$4	$0.20	$20	1M	n/a
Kimi K3	$3	$0.30	$15	1M	#13
Kimi K3.1 Pro	$2	$0.20	$10	1M	#6
Kimi K2.7 Code Highspeed	$1.90	$0.38	$8	262K	n/a
Kimi K2.7 Code	$0.95	$0.19	$4	262K	#35
Kimi K2.6	$0.95	$0.16	$4	262K	#31

USD per 1M tokens, official list price, checked 2026-10-29. Code Arena rank is among the models in our price table.

Kimi K3.1 Pro or Kimi K3?

K3.1 Pro. It lists at $2 in and $10 out against $3 and $15 for K3, and it ranks higher on Code Arena (#6 vs #13). A medium month costs $90.00 on K3.1 Pro and $135 on K3. Moonshot calls K3 the previous flagship, so there is no price reason to stay on it.

What Highspeed costs you

Kimi K3.1 Pro Highspeed is the same model as K3.1 Pro at about 2 to 3 times the speed, per Moonshot. Input and output are both 2x the standard price, while cached input stays at $0.20. A medium month is $180 against $90.00. Pay it only if a person is waiting on each answer; batch jobs should use the standard model.

Cache writes are billed on Kimi

Cache reads are cheap ($0.20 per 1M on K3.1 Pro), but Moonshot also charges to write the cache: $2 per 1M for a 5-minute write and $4 for a 1-hour write on K3.1 Pro. If your prompts rarely repeat inside the window, caching can cost more than it saves.

The K2 series is on its way out

Kimi K2.7 Code and Kimi K2.6 list at $0.95 in and $4 out, which looks cheap. Moonshot says it closes both to new sign-ups from Oct 31, 2026 and plans to retire the K2 series by the end of December 2026. Don't start a new build on them. K2.7 Code Highspeed ($8 out, about 180 tokens per second) is in the same position.

What a month costs

List price, no discounts, our math. Medium is 20M input and 5M output tokens a month; heavy is 200M and 50M.

Model	Medium month	Heavy month
Kimi K2.7 Code	$39.00	$390
Kimi K2.6	$39.00	$390
Kimi K2.7 Code Highspeed	$78.00	$780
Kimi K3.1 Pro	$90.00	$900
Kimi K3	$135	$1,350
Kimi K3.1 Pro Highspeed	$180	$1,800

Kimi against the models people compare it with

Kimi K3.1 Pro vs Claude Opus 5.5: 2x cheaper blended, #6 vs #1 on Code Arena.
Kimi K3.1 Pro vs DeepSeek V4 Pro: $4 vs $1.98 blended.
Kimi K3.1 Pro vs GPT-6.1 Sol: the same list price ($2 / $10), Sol ranks #4 on Code Arena.
Kimi K3.1 Pro vs Qwen3.8 Max and vs GLM-5.3.

Sources

FAQ

How much does the Kimi API cost?

Kimi K3.1 Pro costs $2 per 1M input tokens, $0.20 cached and $10 output. The whole Kimi lineup runs from $0.95 to $4 input, as of 2026-10-29.

What is the context window of Kimi K3.1 Pro?

About 1M tokens (1,048,576), and Moonshot says the price is flat across the whole window.

What is Kimi K3.1 Pro Highspeed?

The same model as K3.1 Pro at about 2 to 3 times the speed, for double the price: $4 input and $20 output per 1M tokens.

Is Kimi K3.1 Pro cheaper than Claude Opus 5.5?

Yes. Kimi K3.1 Pro is $4 blended per 1M tokens against $8 for Claude Opus 5.5, about 2x less. Opus ranks #1 on Code Arena and Kimi #6.

Is Kimi K2.6 being retired?

Moonshot says it closes the K2 series to new sign-ups from Oct 31, 2026 and plans to retire it by the end of December 2026.

Does Kimi charge for prompt caching?

Cache reads are discounted ($0.20 per 1M on K3.1 Pro). Cache writes cost $2 per 1M for 5 minutes or $4 for 1 hour.