Kimi API

What it does

Kimi API is Moonshot AI's developer platform for calling its Kimi models. You send text, images or video through the Chat Completions, Responses or Messages API formats and receive generated text, JSON or tool calls. The newest model is Kimi K3.1 Pro, with a 1M-token context window and native visual understanding; a Highspeed variant trades higher price for faster output.

Developers create an API key, point an OpenAI or Anthropic SDK (or plain HTTP) at the Moonshot base URL, and pick a model. The platform also offers official tools such as web search, code execution, file analysis for Excel and CSV, memory and URL fetching. Typical uses listed include coding agents, research and reasoning, contract review and conversation analysis. Context caching is billed automatically for repeated request prefixes.

It suits developers and teams building applications on a hosted model API, not people wanting a chat app; the Kimi membership and Kimi Code are separate products. Access needs a platform account and API key, with billing per million tokens and no subscription. The K2 series will not be offered to newly registered users from October 31, 2026 and is planned for retirement by the end of December 2026. Rate limits rise with cumulative spend.

What it can do

• Chat, Responses and Messages API access to Kimi K3.1 Pro and K3
• Text, image and video input
• Tool calling, JSON mode and streaming output
• Reasoning effort control on K3.1 Pro
• Official tools including web search, code runner, Excel/CSV analysis and fetch
• Automatic context caching of repeated prompt prefixes

Works with

• OpenAI SDK
• Anthropic SDK

Example workflows

Review a long codebase with a coding agent

Input: repository files and a refactoring request. Steps: create an API key, point an agent tool or SDK at kimi-k3.1-pro, send the code as context and enable tool calls. Output: proposed edits and explanations. Repeated prefixes may be cached.

Analyze a spreadsheet with web context

Input: an Excel or CSV file and a research question. Steps: extract the file content, call the model with the Excel and Web Search tools enabled, and request JSON output. Output: structured findings with cited sources.

Before you choose

• Billing is separate input, output and cache charges per million tokens, so long prompts and large outputs drive cost.
• K2-series models are scheduled for retirement by the end of December 2026; new projects should target K3-series models.
• Rate limits depend on cumulative spend tiers; enterprise needs go through a separate solutions offering.

Listed under

• Coding
• AI agents
• Conversation analysis
• Developers
• Developer tools
• Enterprise

Get started

Sign in to the platform, create an API key under API Keys, then set the base URL.

1. Sign in to the Kimi API Platform and create an API key under API Keys.
2. Store the key in an environment variable such as MOONSHOT_API_KEY.
3. Set base_url to https://api.moonshot.ai/v1 in the OpenAI SDK and call model kimi-k3.1-pro.

Pay-as-you-go with no subscription; prices per 1,000,000 tokens exclude taxes. K3 cache writes are billed by TTL; cached input is billed at the cache-hit price. File APIs are temporarily free.

Source: platform.kimi.ai/docs/pricing/chat, checked 29 Oct 2026. Prices can change; check before you buy.

Sources for this overview

Product pages, documentation, and pricing checked 29 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Kimi API Platform ↗platform.kimi.ai
• Model Inference Pricing - Kimi API Platform ↗platform.kimi.ai
• Quickstart - Kimi API Platform ↗platform.kimi.ai

Company website

platform.kimi.ai: product, pricing and policy pages

YC directory mirror

Batch and company name

Human review

Checked by a person

knowledge_base

Last checked 29 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.