Kimi K3 | EmpirioLabs AI Docs

Moonshot AI · Text Generation

POST /v1/chat/completions

Kimi K3 is Moonshot’s flagship reasoning model with a 1M token context, always-on thinking, native web search, and text, image, and video inputs.

At a glance

Field	Value
Model id	`kimi-k3`
Model release date	2026-07-15
Input modalities	Text, Image, Video
Output modalities	Text
Context window	1M
Weight precision	-
Max output tokens	131,072
Region	International
Features	reasoning, function_calling, multimodal, agentic_coding, web_search
Native inference	No
New	Yes
Structured output	JSON Schema
Supported endpoints	`POST /v1/chat/completions`, `POST /v1/responses`, `POST /v1/messages`, `POST /v1beta/models/kimi-k3:generateContent`
Alternate model ids	`moonshotai/kimi-k3`

Pricing

Charge	Spec	Rate
Input	per 1M prompt tokens	$3.00
Output	per 1M generated tokens	$15.00
Web search	per call when invoked	$0.015

Example request

$ curl https://api.empiriolabs.ai/v1/chat/completions \
>   -H 'Authorization: Bearer $EMPIRIOLABS_API_KEY' \
>   -H 'Content-Type: application/json' \
>   -d '{"model": "kimi-k3", "messages": [{"role":"user","content":"Hello"}]}'

Parameters

Parameter	Type	Required	Default	Description
`max_tokens`	number	no	`32768`	Maximum output tokens. Reasoning tokens count toward this limit. · Range: 1 – 1048576
`stop`	string	no	-	Up to 4 strings where the model will stop generating further tokens.
`reasoning_effort`	enum	no	`"max"`	Kimi K3 reasoning effort. Thinking is always on. Higher effort spends more reasoning tokens before answering. Max is recommended and the default. · Allowed: `low`, `medium`, `high`, `max`
`tool_web_search`	boolean	no	false	Search the web for real-time information. Adds $0.015 to the request cost for each invoked web search call.
`response_format`	enum	no	-	Constrain the output to JSON. Use JSON mode for any valid JSON object, or JSON schema to force output that matches a schema you provide.

Notes

Supports text, image, and video inputs with a 1M token context, function calling, JSON schema structured output, and built-in web search at $0.015 per invoked call. Thinking is always on; reasoning depth is tunable with the reasoning_effort control (max is recommended) and reasoning tokens are billed as output tokens. Temperature and other sampling settings are fixed by the model service. Multi-step function calling through the API must replay the assistant message with its reasoning_content field intact.

Per-tool billing (usage.tool_usage)

When this model invokes built-in tools inside a single request, the response carries a normalized usage.tool_usage map alongside the token counts. Tool counts are already factored into cost_usd and are surfaced for transparency.

Machine-readable schema: GET https://api.empiriolabs.ai/v1/models/kimi-k3.