Can I self-host Kimi?

Moonshot released open weights for its flagship model, so it can be self-hosted as well as used through the API.

Does Kimi charge more for long context?

Moonshot lists a flat rate across the flagship's full context window, with no tiering by length.

Kimi (Moonshot) API pricing

For technical teams

Kimi (Moonshot) is priced per use rather than per month; the rate for each option is in the table below. Checked Nov 10, 2026 against the official pricing page.

Moonshot's Kimi models — strong agentic performance, priced per token.

Visit Kimi (Moonshot) Report a price change

Our take

Score 3.5 out of 5How we score

Moonshot's Kimi models are strong at agentic and coding work, priced per token with a large context window on the flagship. They are a good alternative to the frontier labs for developers who want open-weight options too.

Plans and prices

Kimi (Moonshot) plans, checked Nov 10, 2026. Official pricing page Prices are indicative and may change. Always confirm on the provider's site.
Plan	Price	What you get
K3.1 Pro	Usage-basedNEW model. Input $2.00, cached $0.20, output $10.00. Cache writes $2-4.	1,048,576-token context window New flagship-tier model Context caching
K3.1 Pro HighSpeed	Usage-basedNEW model. Input $4.00, cached $0.20, output $20.00. Cache writes $4 (5-min) / $8 (1-hour).	1,048,576-token context window Low-latency variant
K3 (flagship)	Usage-basedInput (cache miss) $3.00, Input (cache hit) $0.30, Output $15.00. Cache write $3.00 (5min TTL) or $6.00 (1h TTL) — newly confirmed detail. Flat rate across the full 1,048,576-token context window.	1,048,576-token context window; min $1 top-up Batch inference at 60% of standard price
K2.7 CodeOur pick	Usage-basedInput $0.95 (cache hit $0.19), output $4.00 — unchanged; no longer available to NEW users as of 2026-10-31; full K2 series retirement planned end of Dec 2026	262,144-token context window
K2.7 Code HighSpeed	Usage-basedInput $1.90, output $8.00 — unchanged; same K2-retirement timeline	262,144-token context window
K2.6	Usage-basedInput $0.95, output $4.00 — unchanged; same K2-retirement timeline	262,144-token context window

Pick this tier: K2.7 Code

K2.7 Code is the cost-effective choice for coding agents; move to K3 only when you need its larger context or reasoning.

What works

Strong agentic and coding performance
Very large context window on the flagship
Automatic context caching
Open-weight release available for self-hosting

What to watch

Flagship output tokens are not cheap
Older models carry sunset notices
A small minimum top-up is needed to start

Skip it if

You are not building with an API
You need the lowest cost per token
You want a model with a long, stable support life

Best for

Improve coding / developmentResearch competitors

Kimi (Moonshot) pricing questions

How much does Kimi (Moonshot) cost in 2026?: Kimi (Moonshot) is priced per use rather than per month; the rate for each option is in the table below. Checked Nov 10, 2026 against the official pricing page.
Can I self-host Kimi?: Moonshot released open weights for its flagship model, so it can be self-hosted as well as used through the API.
Does Kimi charge more for long context?: Moonshot lists a flat rate across the flagship's full context window, with no tiering by length.

Reviewed Sep 11, 2026 by the Dapols team. Report a price change

Our evidence on this tool

Catalog status: Recommended — can be a primary pick in a plan
Price last checked: Nov 10, 2026
Evidence confidence: Medium — verified, but the vendor changes plans often

Re-checked in batches. We date every check and never take payment for placement. How we rank tools

Is Kimi (Moonshot) right for your business?

Answer a few questions and see whether Kimi (Moonshot) fits your stack and budget.

Find my plan