> ## Documentation Index
> Fetch the complete documentation index at: https://docs.evolink.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# MiniMax-M3 - OpenAI-Compatible API
> - Use the OpenAI Chat Completions protocol to call the MiniMax-M3 model
- **Multi-turn conversation**: Supports single-turn or multi-turn contextual dialogue
- **System prompts**: Customize AI role and behavior via `role=system` messages
- **Multimodal input**: `content` supports mixed text + image / video
- **Thinking mode**: Controlled via `thinking.type`; thinking content is returned via `reasoning_content`
- **Streaming output**: Supports SSE streaming responses
- **Tool calling**: Supports Function Calling
**BaseURL**: The default BaseURL is `https://direct.evolink.ai`, which has better support for text models and long-lived connections. `https://api.evolink.ai` is the primary endpoint for multimodal services and serves as a fallback address for text models.
## OpenAPI
````yaml /en/api-manual/language-series/minimax-m3/minimax-m3-chat.json POST /v1/chat/completions
openapi: 3.1.0
info:
title: MiniMax-M3 Complete API Reference (OpenAI-Compatible)
description: >-
Complete API reference for the MiniMax-M3 chat interface.
**Model Capabilities**:
- Maximum output: **524,288 tokens** (512K), recommended **131,072 tokens**
(128K)
- Thinking mode: Controlled via the `thinking` field, defaults to `adaptive`
(model adaptively decides whether to engage in deep thinking)
- Multimodal input: Supports text, images, and video
- Tool calling: Supports Function Calling
- Streaming output: Supports SSE streaming responses
license:
name: MIT
version: 1.0.0
servers:
- url: https://direct.evolink.ai
description: Production (Recommended)
- url: https://api.evolink.ai
description: Alternative URL
security:
- bearerAuth: []
tags:
- name: Chat Completion
description: AI chat completion related endpoints
paths:
/v1/chat/completions:
post:
tags:
- Chat Completion
summary: MiniMax-M3 Chat Interface (OpenAI-Compatible)
description: >-
- Use the OpenAI Chat Completions protocol to call the MiniMax-M3 model
- **Multi-turn conversation**: Supports single-turn or multi-turn
contextual dialogue
- **System prompts**: Customize AI role and behavior via `role=system`
messages
- **Multimodal input**: `content` supports mixed text + image / video
- **Thinking mode**: Controlled via `thinking.type`; thinking content is
returned via `reasoning_content`
- **Streaming output**: Supports SSE streaming responses
- **Tool calling**: Supports Function Calling
operationId: createChatCompletionMiniMaxM3
requestBody:
required: true
content:
application/json:
schema:
$ref: '#/components/schemas/ChatCompletionRequest'
examples:
simple_text:
summary: Single-turn text conversation
value:
model: MiniMax-M3
messages:
- role: user
content: Please introduce yourself
multi_turn:
summary: Multi-turn conversation (context understanding)
value:
model: MiniMax-M3
messages:
- role: user
content: What is Python?
- role: assistant
content: Python is a high-level programming language...
- role: user
content: What are its advantages?
system_prompt:
summary: Using system prompts
value:
model: MiniMax-M3
messages:
- role: system
content: >-
You are a professional Python programming assistant.
Answer questions concisely.
- role: user
content: How to read a file?
image_understanding:
summary: Image understanding (multimodal input)
value:
model: MiniMax-M3
thinking:
type: adaptive
messages:
- role: user
content:
- type: text
text: What is in this image?
- type: image_url
image_url:
url: https://example.com/image.jpeg
max_completion_tokens: 500
disable_thinking:
summary: Disable thinking mode (direct answer)
value:
model: MiniMax-M3
thinking:
type: disabled
messages:
- role: user
content: What is the capital of France?
tool_calling:
summary: Function Calling (tool invocation)
value:
model: MiniMax-M3
messages:
- role: user
content: Check the weather in Beijing today
tools:
- type: function
function:
name: get_weather
description: Query weather conditions for a specified city
parameters:
type: object
properties:
location:
type: string
description: 'City name, e.g.: Beijing'
required:
- location
streaming:
summary: Streaming output
value:
model: MiniMax-M3
stream: true
stream_options:
include_usage: true
messages:
- role: user
content: Write a short poem about spring
responses:
'200':
description: Chat completion successful
content:
application/json:
schema:
$ref: '#/components/schemas/ChatCompletionResponse'
examples:
default_thinking:
summary: Default (thinking inlined in content within tags)
value:
id: 066b36619b147e326d17053cccdef70f
choices:
- finish_reason: stop
index: 0
message:
content: >-
The user is asking about the capital of France,
which is a common knowledge question. The answer is
Paris.
The capital of France is **Paris**.
role: assistant
name: MiniMax AI
audio_content: ''
created: 1777026807
model: MiniMax-M3
object: chat.completion
usage:
total_tokens: 60
total_characters: 0
prompt_tokens: 7
completion_tokens: 53
prompt_tokens_details:
cached_tokens: 0
input_sensitive: false
output_sensitive: false
input_sensitive_type: 0
output_sensitive_type: 0
base_resp:
status_code: 0
status_msg: ''
tool_calls:
summary: Tool call triggered (finish_reason=tool_calls)
value:
id: 066b13db03518f86c2e3c9b073c04272
choices:
- finish_reason: tool_calls
index: 0
message:
content: >-
The user wants to check the weather in Beijing,
calling the get_weather tool.
Let me check the current weather in Beijing for you.
role: assistant
name: MiniMax AI
tool_calls:
- id: call_function_p4iiqtpnh5bj_1
type: function
function:
name: get_weather
arguments: '{"location": "Beijing"}'
index: 0
audio_content: ''
created: 1777026900
model: MiniMax-M3
object: chat.completion
usage:
total_tokens: 110
total_characters: 0
prompt_tokens: 86
completion_tokens: 24
prompt_tokens_details:
cached_tokens: 0
input_sensitive: false
output_sensitive: false
input_sensitive_type: 0
output_sensitive_type: 0
base_resp:
status_code: 0
status_msg: ''
'400':
description: Invalid request parameters
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
example:
error:
code: 400
message: Invalid request parameters
type: invalid_request_error
'401':
description: Unauthenticated, invalid or expired token
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
example:
error:
code: 401
message: Invalid or expired token
type: authentication_error
'402':
description: Insufficient quota, recharge required
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
example:
error:
code: 402
message: Insufficient quota
type: insufficient_quota_error
'403':
description: Access denied for this model
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
example:
error:
code: 403
message: Access denied for this model
type: permission_error
param: model
'404':
description: Resource not found
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
example:
error:
code: 404
message: Specified model not found
type: not_found_error
param: model
'413':
description: Request body too large
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
example:
error:
code: 413
message: Request body too large
type: request_too_large_error
param: messages
'429':
description: Rate limit exceeded
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
example:
error:
code: 429
message: Rate limit exceeded
type: rate_limit_error
'500':
description: Internal server error
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
example:
error:
code: 500
message: Internal server error
type: internal_server_error
'502':
description: Bad gateway
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
example:
error:
code: 502
message: Bad gateway
type: bad_gateway_error
'503':
description: Service temporarily unavailable
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
example:
error:
code: 503
message: Service temporarily unavailable
type: service_unavailable_error
components:
schemas:
ChatCompletionRequest:
type: object
required:
- model
- messages
properties:
model:
type: string
description: Chat model name
enum:
- MiniMax-M3
example: MiniMax-M3
messages:
type: array
description: >-
List of conversation messages, supports multi-turn dialogue
Messages with different roles have different field structures;
select the corresponding role to view
items:
oneOf:
- $ref: '#/components/schemas/SystemMessage'
- $ref: '#/components/schemas/UserMessage'
- $ref: '#/components/schemas/AssistantRequestMessage'
- $ref: '#/components/schemas/ToolMessage'
discriminator:
propertyName: role
mapping:
system:
$ref: '#/components/schemas/SystemMessage'
user:
$ref: '#/components/schemas/UserMessage'
assistant:
$ref: '#/components/schemas/AssistantRequestMessage'
tool:
$ref: '#/components/schemas/ToolMessage'
minItems: 1
thinking:
type: object
description: >-
Controls deep thinking
**Notes**:
- **Defaults to `adaptive`**: The model adaptively decides whether
to engage in deep thinking based on problem difficulty
- By default, thinking content is inlined in the response `content`
(wrapped in `...` tags); to separate it into a
dedicated field, use `reasoning_split`
properties:
type:
type: string
description: >-
Deep thinking mode
- `adaptive`: Model autonomously decides whether thinking is
needed (recommended default)
- `disabled`: Disable thinking, answer directly
enum:
- disabled
- adaptive
default: adaptive
reasoning_split:
type: boolean
description: >-
Whether to split thinking content into a separate field
- `false` (default): Thinking content is inlined in `content`,
wrapped in `...` tags
- `true`: Thinking content is split into
`choices[].message.reasoning_content` and `reasoning_details`
temperature:
type: number
description: |-
Sampling temperature, controls output randomness
**Notes**:
- Lower values (e.g. 0.2): More deterministic, focused output
- Higher values (e.g. 1.5): More random, creative output
- Range: `[0, 2]`, default 1
minimum: 0
maximum: 2
default: 1
example: 1
top_p:
type: number
description: >-
Nucleus Sampling parameter
**Notes**:
- Controls sampling from tokens with cumulative probability
- e.g. 0.95 means selecting from tokens reaching 95% cumulative
probability
- Range: `[0, 1]`, MiniMax-M3 default 0.95
**Recommendation**: Do not adjust temperature and top_p
simultaneously
minimum: 0
maximum: 1
default: 0.95
example: 0.95
max_completion_tokens:
type: integer
description: >-
Upper limit for generated content length (in tokens)
**Notes**:
- MiniMax-M3 recommended **131,072** (128K), maximum **524,288**
(512K)
- Tokens generated by thinking also count toward this limit
- If generation is interrupted due to `length`, try increasing this
value
minimum: 1
maximum: 524288
example: 131072
stream:
type: boolean
description: >-
Whether to return the response in streaming mode
- `true`: Streaming response, returns content in real-time chunks
via SSE (Server-Sent Events)
- `false`: Wait for complete response before returning (default)
default: false
example: false
stream_options:
type: object
description: |-
Streaming response options
Only effective when `stream=true`
properties:
include_usage:
type: boolean
description: >-
Return an additional chunk containing only `usage` statistics at
the end of the stream
tools:
type: array
description: |-
Tool definition list for Function Calling
Each tool requires a name, description, and parameter schema
items:
$ref: '#/components/schemas/Tool'
max_tokens:
type: integer
description: |-
Legacy generation length limit parameter
**Note**: Deprecated, please use `max_completion_tokens` instead
deprecated: true
minimum: 1
ChatCompletionResponse:
type: object
properties:
id:
type: string
description: Unique identifier for the chat completion
example: 0668a381bdc3c0ded310e27c9a46d16e7
model:
type: string
description: Model name actually used
example: MiniMax-M3
object:
type: string
enum:
- chat.completion
description: Response type
example: chat.completion
created:
type: integer
description: Creation timestamp (Unix seconds)
example: 1777026807
choices:
type: array
description: List of chat completion choices
items:
$ref: '#/components/schemas/Choice'
usage:
$ref: '#/components/schemas/Usage'
input_sensitive:
type: boolean
description: >-
Whether the input content triggered a sensitive word filter. If the
input severely violates policies, the API will return a content
violation error with empty response content
input_sensitive_type:
type: integer
description: >-
Type of sensitive word triggered by input (returned when
`input_sensitive` is true): 1 severe violation; 2 pornography; 3
advertising; 4 prohibited content; 5 abusive language; 6
violence/terrorism; 7 other
output_sensitive:
type: boolean
description: Whether the output content triggered a sensitive word filter
output_sensitive_type:
type: integer
description: Type of sensitive word triggered by output
base_resp:
type: object
description: Status code and error details
properties:
status_code:
type: integer
description: >-
Status code. `0` indicates success; other codes include `1002`
for rate limiting, `1008` for insufficient balance, `1039` for
token limit exceeded, `2013` for parameter error, etc.
status_msg:
type: string
description: Error details
ErrorResponse:
type: object
properties:
error:
type: object
properties:
code:
type: integer
description: HTTP status error code
message:
type: string
description: Error description
type:
type: string
description: Error type
param:
type: string
description: Related parameter name
SystemMessage:
title: System Message
type: object
required:
- role
- content
properties:
role:
type: string
enum:
- system
description: Role identifier, fixed as `system`
content:
type: string
description: System prompt content, used to set the AI's role and behavior
name:
type: string
description: >-
Participant name, used to distinguish different system prompt
sources
UserMessage:
title: User Message
type: object
required:
- role
- content
properties:
role:
type: string
enum:
- user
description: Role identifier, fixed as `user`
content:
description: >-
User message content
**Notes**:
- For plain text, pass a string directly
- For multimodal, pass a content part array supporting `text` /
`image_url` (images, video)
oneOf:
- type: string
- type: array
items:
$ref: '#/components/schemas/ContentPart'
name:
type: string
description: Participant name, used to distinguish different users
AssistantRequestMessage:
title: Assistant Message
type: object
required:
- role
properties:
role:
type: string
enum:
- assistant
description: Role identifier, fixed as `assistant`
content:
type:
- string
- 'null'
description: >-
Assistant message content
**Notes**:
- Used to pass historical assistant replies in multi-turn
conversations
- Can be `null` when `tool_calls` is present
name:
type: string
description: Participant name
tool_calls:
type: array
description: >-
Tool call list
Used to pass historical tool call information in multi-turn
conversations
items:
type: object
properties:
id:
type: string
description: Unique identifier for the tool call
type:
type: string
enum:
- function
function:
type: object
properties:
name:
type: string
description: Name of the called function
arguments:
type: string
description: Function arguments (JSON string)
ToolMessage:
title: Tool Message
type: object
required:
- role
- content
- tool_call_id
properties:
role:
type: string
enum:
- tool
description: Role identifier, fixed as `tool`
content:
type: string
description: Content of the tool call result
tool_call_id:
type: string
description: >-
Tool call ID
Corresponds to the `id` field returned in `tool_calls` of the
assistant message
Tool:
type: object
required:
- type
- function
properties:
type:
type: string
enum:
- function
description: Tool type, currently only `function` is supported
function:
type: object
required:
- name
- parameters
properties:
name:
type: string
description: >-
Name of the function to call
**Notes**:
- Must consist of a-z, A-Z, 0-9 characters, or include
underscores and hyphens
- Maximum length is 64 characters
description:
type: string
description: >-
Function description, helps the model understand when and how to
call this function
parameters:
type: object
description: Function input parameters, described as a JSON Schema object
Choice:
type: object
properties:
index:
type: integer
description: Choice index
example: 0
message:
$ref: '#/components/schemas/AssistantMessage'
finish_reason:
type: string
description: |-
Finish reason
- `stop`: Natural completion or stop sequence triggered
- `length`: Reached maximum token limit
- `content_filter`: Content filtered by safety policy
- `tool_calls`: Model invoked a tool
enum:
- stop
- length
- content_filter
- tool_calls
example: stop
Usage:
type: object
description: Token usage statistics
properties:
total_tokens:
type: integer
description: Total number of tokens consumed
example: 1604
total_characters:
type: integer
description: Total number of characters consumed
example: 0
prompt_tokens:
type: integer
description: Number of tokens in the input
example: 1365
completion_tokens:
type: integer
description: Number of tokens in the output
example: 239
prompt_tokens_details:
type: object
description: Detailed breakdown of input tokens
properties:
cached_tokens:
type: integer
description: Number of input tokens that hit cache
example: 114
ContentPart:
type: object
description: >-
Multimodal content part. Declare the type via `type` and fill in only
the fields matching that type
properties:
type:
type: string
enum:
- text
- image_url
- video_url
description: |-
Content type
- `text`: Text block
- `image_url`: Image input
- `video_url`: Video input
text:
type: string
description: Text content when `type=text`
image_url:
type: object
description: >-
Image input (when `type=image_url`). Maximum 10 MB per image,
supports JPEG / PNG / GIF / WEBP
required:
- url
properties:
url:
type: string
description: Image URL, or a base64 data URL starting with `data:`
detail:
type: string
description: >-
Controls image parsing resolution
- `low`: Low precision, consumes fewer tokens
- `default`: Default precision
- `high`: High precision, recognizes more detail but consumes
more tokens
enum:
- low
- default
- high
default: default
max_long_side_pixel:
type: integer
description: Maximum pixel limit for the longest side of the image
minimum: 1
video_url:
type: object
description: >-
Video input (when `type=video_url`). Maximum 50 MB for URL / Base64
video, supports MP4 / AVI / MOV / MKV
required:
- url
properties:
url:
type: string
description: >-
Video URL, a base64 data URL starting with `data:`, or a file
reference in the format `mm_file://{file_id}`
detail:
type: string
description: Controls video frame extraction resolution
enum:
- low
- default
- high
default: default
fps:
type: number
description: >-
Video frame extraction rate, default 1, range [0.2, 5]
- Higher values: More sensitive to visual changes, higher token
cost and slower speed
- Lower values: Lower token cost and faster speed, but less
sensitive to visual changes
minimum: 0.2
maximum: 5
default: 1
max_long_side_pixel:
type: integer
description: Maximum pixel limit for the longest side of video frames
minimum: 1
AssistantMessage:
type: object
properties:
role:
type: string
description: Role of the message sender
enum:
- assistant
example: assistant
content:
type:
- string
- 'null'
description: >-
AI response message content (can be `null` when a tool call is
triggered)
example: >-
Hello! I'm MiniMax-M3, and I can help you with conversation,
writing, reasoning, coding, and many other tasks.
reasoning_content:
type: string
description: >-
Thinking content. Only returned when `reasoning_split=true` is set
in the request; by default, thinking content is inlined in `content`
within `...` tags
example: Let me analyze this problem...
reasoning_details:
type: array
description: >-
Structured thinking content. Only returned when
`reasoning_split=true` is set in the request
items:
type: object
properties:
type:
type: string
description: Thinking content type, e.g. `reasoning.text`
id:
type: string
description: Identifier for the thinking segment
format:
type: string
description: >-
Format identifier for the thinking content, e.g.
`MiniMax-response-v1`
index:
type: integer
description: Sequential index of the thinking segment
text:
type: string
description: Thinking text of this segment
name:
type: string
description: Sender name
audio_content:
type: string
description: Audio content (empty string for text-only conversation scenarios)
tool_calls:
type: array
description: Tool call list (returned when the model decides to invoke a tool)
items:
type: object
properties:
id:
type: string
description: Unique identifier for the tool call
type:
type: string
enum:
- function
function:
type: object
properties:
name:
type: string
description: Name of the called function
arguments:
type: string
description: Function arguments (JSON string)
index:
type: integer
description: Tool call index
securitySchemes:
bearerAuth:
type: http
scheme: bearer
description: >-
##All APIs require Bearer Token authentication##
**Get API Key:**
Visit [API Key Management Page](https://evolink.ai/dashboard/keys) to
get your API Key
**Add to request header:**
```
Authorization: Bearer YOUR_API_KEY
```
````