China AI Hub China AI Hub

API / moonshot

moonshot API

Provider: moonshot-ai · Type: official

moonshot is an API by moonshot-ai, served at https://api.moonshot.ai/v1 for official models, exposing Kimi K2.6, Kimi K2.7 Code, Kimi K3.1 Pro Highspeed, Kimi K3.1 Pro, Kimi K3, Kimi K2.7 Code Highspeed.

moonshot API
Image: AI-generated illustration (Seedream)
Last verified: · Review cadence: 90 days · Next review:
Key facts for moonshot API
Key factValue
Endpointhttps://api.moonshot.ai/v1
AuthenticationBearer API key (MOONSHOT_API_KEY), managed in the platform console
StreamingYes
Function callingYes
Tool callingYes
Structured outputYes
VisionYes
AudioUnknown
Regions—
Cloud providersMoonshot AI Platform
Documentationhttps://platform.kimi.ai/docs

Context limits

Context limits for moonshot API
ModelInput limitOutput limit
kimi-k3 1,048,576 1,048,576
kimi-k2.7-code 262,144 —
kimi-k2.7-code-highspeed 262,144 —
kimi-k2.6 262,144 —

What it is. The Moonshot AI (Kimi) API serves the Kimi K-series at api.moonshot.ai, exposing OpenAI-compatible Chat Completions and Responses plus an Anthropic-compatible Messages endpoint. Why it matters. It is the developer path to Kimi K3’s 1M-token input and output context — the database’s only full-window output ceiling. Key characteristics. Separate reasoning_content and content stream deltas; K3 uses reasoning_effort (low/high/max) while K2.x uses thinking; JSON mode / structured output, automatic context caching, file upload (text/image/video). What a professional should know. Temperature and top_p are fixed for K3 and K2.x (not tunable), and rate limits/regions are published behind a JS-rendered page that could not be extracted into structured data.

The Moonshot AI (Kimi) API serves the Kimi K-series at api.moonshot.ai. It exposes OpenAI-compatible Chat Completions (/v1/chat/completions), OpenAI Responses (/v1/responses) and an Anthropic-compatible Messages endpoint (/anthropic/v1/messages), plus files, batches, tokenizer and web-search tools.

Streaming is SSE with separate reasoning_content and content deltas. kimi-k3 uses top-level reasoning_effort (low/high/max, default max); K2.x models use the thinking parameter. JSON mode / structured output, automatic context caching and file upload (text/image/video) are supported. Temperature and top_p are fixed for K3 and K2.x. Serving regions are not stated in the fetched docs.

Why it matters

The Kimi API matters as the long-context-generation path: Kimi K3’s 1M-token output ceiling — unique in the database — is only reachable through this API, making it the reference surface for long-form generation and whole-repo rewriting. The cost trade-off is the premium: $3.00/$15.00 per 1M tokens for kimi-k3, the highest input/output pricing among the non-Doubao flagships. China AI Hub analysis indicates the API’s structural role is the long-output specialist, with fixed sampling parameters (temperature/top_p) as the notable constraint for teams that normally tune generation.

Labels used above: Official fact (from the Kimi API overview and Chat Completions spec), Vendor-reported claim (pricing and capability statements by Moonshot AI), and China AI Hub analysis (our synthesis, always introduced as such).

Sources

Data interpretation

Evidence types for moonshot API
FieldValueEvidence type
Endpointhttps://api.moonshot.ai/v1Official
AuthenticationBearer API key (MOONSHOT_API_KEY), managed in the platform consoleOfficial
CapabilitiesStreaming, Function calling, Tool calling, Structured output, VisionOfficial
Models served6 models in the databaseChina AI Hub analysis

Evidence types: Official = vendor documentation or official pages. China AI Hub analysis = derived from the database itself. See the sourcing policy.

What is the moonshot API endpoint?

The moonshot API endpoint is https://api.moonshot.ai/v1.

How do you authenticate with the moonshot API?

Bearer API key (MOONSHOT_API_KEY), managed in the platform console.

What are the context limits of the moonshot API?

kimi-k3: 1,048,576 in / 1,048,576 out; kimi-k2.7-code: 262,144 in / — out; kimi-k2.7-code-highspeed: 262,144 in / — out; kimi-k2.6: 262,144 in / — out.

Does the moonshot API support function calling?

Yes — the moonshot API supports function calling.

Where does China AI Hub get its moonshot API data?

From 2 sources, last verified 2026-09-20.