> ## Documentation Index
> Fetch the complete documentation index at: https://evolink.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.
# Qwen3.8-Max-Preview - Full Parameter Documentation
> > 🚧 **This model is not yet available, stay tuned**
- Call Qwen3.8-Max-Preview using the OpenAI Chat Completions protocol
- **Multi-turn conversation**: supports single-turn or multi-turn contextual conversations
- **System prompt**: set the AI's role and behavior via a `role=system` message
- **Multimodal input**: pass a content part array in `content`, supporting `text` / `image_url` / `input_audio` / `video_url`
- **Context caching**: add `cache_control` on a content part to declare explicit caching; see `usage.prompt_tokens_details` in the response for hit details
- **Thinking mode**: enabled with `enable_thinking=true`, thinking content returned through `reasoning_content`
- **Streaming output**: when `stream=true`, returned chunk by chunk via SSE
**BaseURL note**: The default BaseURL is `https://direct.evolink.ai`, which offers better support for text models and long connections. `https://api.evolink.ai` is the primary multimodal endpoint; use it when your request includes image / audio / video input.
## OpenAPI
````yaml en/api-manual/language-series/qwen3.8-max-preview/qwen3.8-max-preview-chat.json POST /v1/chat/completions
openapi: 3.1.0
info:
title: Qwen3.8-Max-Preview Full Parameter Documentation (OpenAI-Compatible)
description: >-
API reference for the Tongyi Qwen3.8-Max-Preview chat API (OpenAI Chat
Completions compatible).
**Model capabilities**:
- Context length: **1,000,000 tokens** (1M)
- Thinking mode: enabled via `enable_thinking`, thinking content returned
through `reasoning_content`
- Multimodal input: `content` supports mixed text + image / audio / video
- Context caching: supports explicit caching (`cache_control`) and implicit
caching, cache hits significantly reduce cost
- Tool calling: supports Function Calling
- Streaming output: supports SSE streaming
license:
name: MIT
version: 1.0.0
servers:
- url: https://direct.evolink.ai
description: >-
Production environment (recommended, better long-connection support for
text models)
- url: https://api.evolink.ai
description: >-
Primary multimodal endpoint (use when input includes image / audio /
video)
security:
- bearerAuth: []
tags: []
paths:
/v1/chat/completions:
post:
summary: Qwen3.8-Max-Preview Chat API (OpenAI-Compatible)
description: >-
> 🚧 **This model is not yet available, stay tuned**
- Call Qwen3.8-Max-Preview using the OpenAI Chat Completions protocol
- **Multi-turn conversation**: supports single-turn or multi-turn
contextual conversations
- **System prompt**: set the AI's role and behavior via a `role=system`
message
- **Multimodal input**: pass a content part array in `content`,
supporting `text` / `image_url` / `input_audio` / `video_url`
- **Context caching**: add `cache_control` on a content part to declare
explicit caching; see `usage.prompt_tokens_details` in the response for
hit details
- **Thinking mode**: enabled with `enable_thinking=true`, thinking
content returned through `reasoning_content`
- **Streaming output**: when `stream=true`, returned chunk by chunk via
SSE
operationId: createChatCompletionQwen38MaxPreview
requestBody:
required: true
content:
application/json:
schema:
$ref: '#/components/schemas/ChatCompletionRequest'
examples:
basic:
summary: Basic text conversation
value:
model: qwen3.8-max-preview
messages:
- role: system
content: You are a helpful assistant.
- role: user
content: Hello, please introduce yourself
multimodal:
summary: Multimodal input (image)
value:
model: qwen3.8-max-preview
messages:
- role: user
content:
- type: text
text: Describe the content of this image
- type: image_url
image_url:
url: https://example.com/image.jpg
explicit_cache:
summary: Explicit caching (cache_control)
value:
model: qwen3.8-max-preview
messages:
- role: system
content:
- type: text
text:
cache_control:
type: ephemeral
- role: user
content: Answer the question based on the context above
responses:
'200':
description: Conversation generated successfully
content:
application/json:
schema:
$ref: '#/components/schemas/ChatCompletionResponse'
'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 balance
content:
application/json:
schema:
$ref: '#/components/schemas/ErrorResponse'
example:
error:
code: 402
message: Insufficient credits
type: insufficient_quota
'429':
description: Too many requests (rate limited)
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: server_error
components:
schemas:
ChatCompletionRequest:
type: object
required:
- model
- messages
properties:
model:
type: string
description: Chat model name
enum:
- qwen3.8-max-preview
example: qwen3.8-max-preview
messages:
type: array
description: >-
List of conversation messages, supporting multi-turn conversations.
Messages from different roles (system / user / assistant / tool)
have different field structures; select the corresponding role to
view details.
items:
oneOf:
- $ref: '#/components/schemas/SystemMessage'
- $ref: '#/components/schemas/UserMessage'
- $ref: '#/components/schemas/AssistantRequestMessage'
- $ref: '#/components/schemas/ToolMessage'
enable_thinking:
type: boolean
description: >-
Whether to enable deep thinking
- `true`: the model outputs its thinking process, returned through
`reasoning_content`
- `false` (default): does not output the thinking process
> Note: some models require this to be explicitly set to `true` to
return thinking content in non-streaming calls.
default: false
temperature:
type: number
description: >-
Sampling temperature, controls output randomness. Lower values are
more deterministic, higher values more diverse. Range `[0, 2]`. It
is recommended not to adjust temperature and top_p at the same time.
minimum: 0
maximum: 2
top_p:
type: number
description: >-
Nucleus Sampling parameter, samples from the tokens within the top
cumulative probability. Range `(0, 1]`. It is recommended not to
adjust temperature and top_p at the same time.
minimum: 0
maximum: 1
max_completion_tokens:
type: integer
description: >-
Upper limit on the length of generated content (in tokens),
**including the chain of thought and the answer**. Recommended for
thinking models. The default and maximum values are both the model's
maximum output length; once exceeded, generation stops early with
`finish_reason=length`.
max_tokens:
type: integer
description: >-
Legacy generation length limit parameter.
> **Deprecated**: new integrations should use
`max_completion_tokens` instead. This parameter only limits the
answer portion (excluding the chain of thought).
stream:
type: boolean
description: |-
Whether to return the response in a streaming manner.
- `true`: returned chunk by chunk via SSE (Server-Sent Events)
- `false` (default): returns the full response at once
default: false
stream_options:
type: object
description: Streaming response options, effective only when `stream=true`.
properties:
include_usage:
type: boolean
description: >-
When `true`, returns an additional data chunk containing only
`usage` before the stream ends.
tools:
type: array
description: >-
List of tool definitions for Function Calling. Each tool must define
a name, description, and parameter schema.
items:
$ref: '#/components/schemas/Tool'
ChatCompletionResponse:
type: object
properties:
id:
type: string
description: Unique identifier for this conversation
example: chatcmpl-xxxxxxxx
object:
type: string
example: chat.completion
created:
type: integer
description: Creation time (Unix seconds)
example: 1735120033
model:
type: string
example: qwen3.8-max-preview
choices:
type: array
items:
$ref: '#/components/schemas/Choice'
usage:
$ref: '#/components/schemas/Usage'
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: Name of the related parameter
SystemMessage:
type: object
title: System Message
required:
- role
- content
properties:
role:
type: string
enum:
- system
description: Message role, fixed to `system`
content:
description: >-
System instruction. Pass a string for plain text; pass a content
part array when declaring explicit caching.
oneOf:
- type: string
- type: array
items:
$ref: '#/components/schemas/ContentPart'
UserMessage:
type: object
title: User Message
required:
- role
- content
properties:
role:
type: string
enum:
- user
description: Message role, fixed to `user`
content:
description: >-
User message content.
- Pass a string directly for plain text
- For multimodal input or explicit caching, pass a content part
array supporting `text` / `image_url` / `input_audio` / `video_url`
oneOf:
- type: string
- type: array
items:
$ref: '#/components/schemas/ContentPart'
AssistantRequestMessage:
type: object
title: Assistant Message
required:
- role
properties:
role:
type: string
enum:
- assistant
description: Message role, fixed to `assistant`
content:
description: >-
Assistant's historical reply content. Can be a string or a content
part array.
oneOf:
- type: string
- type: array
items:
$ref: '#/components/schemas/ContentPart'
tool_calls:
type: array
description: >-
List of tool calls initiated by the assistant in the previous turn
(used to backfill Function Calling history).
items:
type: object
ToolMessage:
type: object
title: Tool Message
required:
- role
- content
- tool_call_id
properties:
role:
type: string
enum:
- tool
description: Message role, fixed to `tool` (tool execution result)
content:
description: >-
Tool execution result content. Can be a string or a content part
array.
oneOf:
- type: string
- type: array
items:
$ref: '#/components/schemas/ContentPart'
tool_call_id:
type: string
description: ID of the corresponding tool call
Tool:
type: object
description: Function Calling tool definition.
required:
- type
- function
properties:
type:
type: string
enum:
- function
description: Tool type, fixed to `function`
function:
type: object
required:
- name
properties:
name:
type: string
description: Function name
description:
type: string
description: Description of the function's purpose
parameters:
type: object
description: JSON Schema of the function parameters
Choice:
type: object
properties:
index:
type: integer
example: 0
message:
$ref: '#/components/schemas/AssistantMessage'
finish_reason:
type: string
description: |-
Stop reason
- `stop`: normal completion
- `length`: reached the `max_completion_tokens` limit
- `tool_calls`: a tool call is required
example: stop
Usage:
type: object
description: Token usage statistics.
properties:
prompt_tokens:
type: integer
description: Number of input tokens
example: 3019
completion_tokens:
type: integer
description: Number of output tokens (including thinking)
example: 104
total_tokens:
type: integer
description: Total number of tokens
example: 3123
completion_tokens_details:
type: object
description: Breakdown of output tokens
properties:
reasoning_tokens:
type: integer
description: Number of thinking-process tokens
example: 0
prompt_tokens_details:
type: object
description: Breakdown of input token caching
properties:
cached_tokens:
type: integer
description: Number of input tokens served from cache (cache read)
example: 2048
cache_creation_input_tokens:
type: integer
description: >-
Number of tokens newly written to cache in this request (cache
write, returned for explicit caching)
example: 0
ContentPart:
type: object
description: >-
Multimodal content part; declare the type via `type` and fill in only
the fields matching that type. You can attach `cache_control` on any
part to declare explicit caching.
required:
- type
properties:
type:
type: string
enum:
- text
- image_url
- input_audio
- video_url
description: |-
Content type
- `text`: text block
- `image_url`: image input
- `input_audio`: audio input
- `video_url`: video file input
text:
type: string
description: Text content when `type=text`
image_url:
type: object
description: >-
Image input (when `type=image_url`). 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: |-
Image parsing precision
- `low`: low precision, consumes fewer tokens
- `high`: high precision, finer recognition
- `auto` (default): decided automatically by the model
enum:
- low
- high
- auto
default: auto
input_audio:
type: object
description: Audio input (when `type=input_audio`).
required:
- data
- format
properties:
data:
type: string
description: Audio URL, or a base64 data URL starting with `data:`
format:
type: string
description: Audio format, such as `mp3`, `wav`, etc.
video_url:
type: object
description: Video file input (when `type=video_url`).
required:
- url
properties:
url:
type: string
description: Video file URL, or a base64 data URL starting with `data:`
cache_control:
type: object
description: >-
Explicit cache marker. The system uses this marker's position as the
endpoint and traces backward to create / hit a cache block.
- Minimum cacheable length **1024 tokens**, valid for **5 minutes**
(reset on hit)
- Up to **4** cache markers per request
- Mutually exclusive with implicit caching; only one applies per
request
See `usage.prompt_tokens_details` in the response for hit / creation
details.
required:
- type
properties:
type:
type: string
enum:
- ephemeral
description: Cache type, fixed to `ephemeral` (5-minute temporary cache)
AssistantMessage:
type: object
description: Assistant reply message.
properties:
role:
type: string
example: assistant
content:
type: string
description: Reply body
example: Hello! I am Tongyi Qwen3.8-Max-Preview.
reasoning_content:
type: string
description: Thinking process content (returned when `enable_thinking=true`)
tool_calls:
type: array
description: List of tool calls initiated by the model
items:
type: object
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
```
````