KimiK3 Max logoKimiK3 Max

Kimi K3 FAQ

The most-asked questions about Moonshot AI's Kimi K3 model — answered in plain language. Want to try it yourself? Open the Kimi K3 playground →

Quick links: Specs · Benchmarks · Pricing · API · Open Weights · How-To Guide
KimiK3 Max is an independent fan and reference site. It is not affiliated with, endorsed by, or operated by Moonshot AI. “Kimi” is a trademark of Moonshot AI. All trademarks, model names, and benchmark data belong to their respective owners.

Frequently Asked Questions

What is Kimi K3?

Kimi K3 is Moonshot AI's flagship large language model, released July 16, 2026. It uses a Mixture-of-Experts design with a reported 2.8 trillion total parameters and a 1,000,000-token context window, and Moonshot plans to release it as an open-weight model.

Who makes Kimi K3?

Kimi K3 is developed by Moonshot AI, the Beijing-based lab behind the Kimi assistant. "Kimi" is a trademark of Moonshot AI.

How many parameters does Kimi K3 have?

Moonshot reports roughly 2.8 trillion total parameters using a sparse Mixture-of-Experts design (896 experts, 16 active per token). The exact active-parameter count has not been disclosed.

What is the Kimi K3 context window?

Kimi K3 supports a context window of up to 1,000,000 tokens, enabled by Kimi Delta Attention, which cuts KV-cache memory at long context.

Is Kimi K3 free to use?

The Kimi assistant on kimi.ai offers a free tier for chat. The developer API is paid, billed per token (cache-hit input $0.30, cache-miss input $3.00, output $15.00 per 1M tokens). Self-hosting the open weights has no API fee but requires your own GPU hardware.

How much does the Kimi K3 API cost?

Official API pricing is $0.30 per million cache-hit input tokens, $3.00 per million cache-miss input tokens, and $15.00 per million output tokens. Use our token cost calculator for estimates.

Is Kimi K3 open source / open weights?

Moonshot describes K3 as an open-weight model. The full model weights were scheduled for release on July 27, 2026, under a Modified MIT licence, but verify the final licence terms on the official release before redistribution.

What can Kimi K3 do?

K3 accepts text, image, and video input, keeps reasoning always on with a tunable reasoning_effort control, and is strong at long-context analysis, coding, agentic tool use, and deep research. On the Artificial Analysis Intelligence Index it scores around 57.

How do I access Kimi K3?

Three ways: (1) chat for free at kimi.ai, (2) call the developer API with an API key, or (3) download the open weights and self-host. See our how-to guide and open-weights pages for steps.

Does Kimi K3 support function calling / tools?

Yes. The Kimi K3 API supports tool use (function calling) and agentic workflows, and the chat interface can browse and run deep-research tasks. Exact tool schemas are documented in the official API reference.

What is the difference between Kimi K3 and Kimi K2?

K3 (July 2026) is a major rebuild over K2: it nearly triples total parameters to ~2.8T, introduces Kimi Delta Attention and Attention Residuals, and pushes context to 1M tokens. K2 was the prior-generation model and remains available.

Is KimiK3 Max affiliated with Moonshot AI?

No. KimiK3 Max is an independent fan and reference site. It is not affiliated with, endorsed by, or operated by Moonshot AI.