Moonshot AI Kimi K3 Pricing

Last updated:

Kimi K3 costs $3.00 input / $15.00 output per 1M tokens through Moonshot AI. Cached input is $0.30 per 1M tokens. The official endpoint has a 1,048,576-token context window, while the open-weight release gives teams a separate self-hosting path with a very different infrastructure bill.

Kimi K3 API token prices

Relative priceLowestLowerHigherHighestAutomatic log scale across current offers; input, cache and output are graded separately. Color shows cost, not quality.
Showing 1 of 1 grouped models from 1 offers · USD per 1M tokens
Last synced:

How we verify pricing data and normalize token costs.

Sponsored links may earn us a commission at no extra cost to you. Affiliate status never changes model ordering.

All product names, logos, and brands are property of their respective owners and are used for identification purposes only.

These are Moonshot AI's first-party list prices. Hosted providers can publish different rates, latency, caching, or support terms for the same model family.

Kimi K3 cache-write and token rates

USD per 1M tokens, excluding applicable taxes. Read, write, and uncached remainder are mutually exclusive input buckets.

Billing bucketRate
Uncached input remainder$3.00
Cached input read$0.30
Cache write · 5-minute TTL$3.00
Cache write · 1-hour TTL$6.00
Output$15.00

Pricing data: How we verify →

Default TTL is 5 minutes. Cache hits renew the original TTL free of additional write charges. The route calculator excludes mixed-cache billing. Rates rendered from the maintained pricing API; official rate card and cache billing guide.

Kimi K3 vLLM optimization: faster decode, unchanged API rates

vLLM reports 842 decode tokens/s on 16 GB300 GPUs with synthetic speculative acceptance. Its AgentX 1.8× interactivity comparison uses 16 GPUs versus 8; neither result proves lower customer API prices. The merged path is opt-in and hardware-specific.

Read the vLLM speed and self-hosting cost analysis · vLLM availability API

Managed API versus self-hosting

Moonshot's managed API is the simplest route for variable traffic: you pay for tokens and avoid owning an accelerator pool. Self-hosting exchanges that bill for fixed hardware capacity, serving engineering, monitoring, upgrades, and utilization risk.

The July 29 imec test makes the scale concrete. Its roughly 1.4TB Kimi K3 checkpoint did not fit on an 8×B200 node with enough room left for KV cache. The team moved to 8×B300, estimated the hardware at about 20% more than its GLM-5.2 B200 setup, and measured lower throughput and longer task time—but a 23.9-point task-resolution lead on its 64-task subset.

Read the complete methodology and contamination caveat in our Kimi K3 API versus self-hosting analysis. Use the token cost calculator to price managed volume before comparing it with a continuously rented B300 node.

Open weights do not mean zero cost

Moonshot describes Kimi K3 as a 2.8T-parameter mixture-of-experts model with 104B active parameters, MXFP4 weights, native multimodality, and a one-million-token context window. The weights are downloadable under the Kimi K3 License, but teams still need to review those terms and budget for storage, HBM, KV cache, power or rental, networking, and on-call operations.

If operating a multi-GPU node is not economical, compare Novita's managed Kimi K3 route with Moonshot's first-party API and other hosts.

Affiliate disclosure: we may earn a commission from the sponsored link above. It does not affect the pricing table or analysis.

What the Kimi K3 “LLM City” shows

The independent LLM City visualization maps Kimi K3's tensor shapes into a 3D architecture and an exact-area atlas. It counts 2,779,931,834,976 logical weight positions—consistent with Moonshot's rounded 2.8T total—and assigns every scalar a user-selectable physical tile. At the default 2.5 mm setting, that area is about 17.37 km², equal to a square roughly 4.17 km on each side.

This is an architecture-scale metaphor, not a dump of learned values. The page says its close-up cell numbers are deterministic mock values. Also keep total and active parameters separate: Moonshot says K3 activates about 104B parameters per token, so the 2.8T city is useful for understanding storage and deployment scale, not as a direct measure of per-token compute or API price.

Read our Kimi K3 model-size and LLM City cost analysis for the scale math and buyer implications.

What the four-SSD MacBook result proves

The independent ARGODRIVE Deltafin project reports 1.0015 tokens/s of steady decode for a 512-token raw completion on an M5 Max MacBook Pro with 128GB unified memory. The setup streamed Kimi K3's expert bank from the internal SSD and three external NVMe enclosures. The same package reports about 375 seconds to the first token for a 512-token prompt.

This is a reproducible systems experiment, not a Moonshot performance claim or a price change. Its four-drive ladder, project-specific storage layout, speculative drafter, int8 resident spine, cold-run method, and prompt dependence make the result unsuitable as a universal local-speed estimate. Read the Kimi K3 MacBook Pro cost and latency analysis for the measurement boundaries and managed-API comparison.

Labs status

Kimi K3 is live in the AI Pricing Guru Labs leaderboard. The maintained run from answered 49/49 deterministic tasks correctly with 0 endpoint errors through OpenRouter. It does not measure the GB300 specialized vLLM path.

Labs still does not test the private imec task subset, SGLang configuration, B300 utilization, Deltafin's four-SSD local throughput, specialized vLLM GB300 throughput, vision performance, or LLM City's architectural representation. It measures public-endpoint cost per correct answer on a small text suite. The local replay requirements are documented in our benchmark coverage notes.

Frequently asked questions

How much does the Kimi K3 API cost?

Moonshot AI lists Kimi K3 at $3.00 per 1M uncached input tokens, $0.30 per 1M cached input tokens, and $15.00 per 1M output tokens.

Can Kimi K3 be self-hosted?

Yes. Moonshot AI publishes Kimi K3 as an open-weight 2.8-trillion-parameter model under the Kimi K3 License. The model card lists MXFP4 weights and 104 billion activated parameters, so production hosting still requires substantial accelerator memory.

What hardware does Kimi K3 need?

Moonshot AI does not publish one universal minimum server. In imec’s July 2026 test, roughly 1.4TB of weights did not leave KV-cache headroom on 8×B200, so the team used 8×B300 with 2.3TB total HBM. Quantization, context, concurrency, and serving software change the requirement.

Can Kimi K3 run on a MacBook Pro?

An independent Deltafin benchmark reports about 1 token per second of steady decode on an M5 Max MacBook Pro with 128GB unified memory and four SSDs. It also reports roughly 375 seconds to the first token for a 512-token prompt, so this is an experimental local route rather than an API-speed replacement.

Is Kimi K3 included in AI Pricing Guru Labs?

Yes. The current managed-route run answered 49 of 49 deterministic tasks correctly, with 0 endpoint errors. It does not measure specialized vLLM GPU throughput.

Official and benchmark sources