MaaS_Qwen

Capability Map

qwen3.8-max

Capability	Support Status	Capability	Support Status
Input Modality	Image　Text　Video	Output Modality	Text
Model Playground	Supported	Function Calling	Supported
Structured Output	Supported	Web Search	Supported
Prefix Continuation	Supported	Context Caching	Supported
Batch Inference	Supported	Model Fine-tuning	Not Supported

Parameter	Value	Parameter	Value
Max Input Length	991808	Max Output Length	131072
Max Input Length (Thinking Mode)	983616	Max Output Length (Thinking Mode)	131072
Context Length	1000000	Max Chain-of-Thought Length	262144

qwen3.7-max

Capability	Support Status	Capability	Support Status
Input Modality	Image　Text　Video	Output Modality	Text
Model Playground	Supported	Function Calling	Supported
Structured Output	Supported	Web Search	Supported
Prefix Continuation	Supported	Context Caching	Supported
Batch Inference	Supported	Model Fine-tuning	Not Supported

Parameter	Value	Parameter	Value
Max Input Length	991808	Max Output Length	131072
Context Length	1000000	Max Input Length (Thinking Mode)	983616
Max Output Length (Thinking Mode)	131072	Max Chain-of-Thought Length	262144

qwen3.5-plus

Capability	Support Status	Capability	Support Status
Input Modality	Image　Text　Video	Output Modality	Text
Model Playground	Supported	Function Calling	Supported
Structured Output	Supported	Web Search	Supported
Prefix Continuation	Supported	Context Caching	Supported
Batch Inference	Supported	Model Fine-tuning	Not Supported

Context Limits

Parameter	Value	Parameter	Value
Max Input Length	991808	Max Output Length	65536
Context Length	1000000	Max Input Length (Thinking Mode)	983616
Max Output Length (Thinking Mode)	65536	Max Chain-of-Thought Length	81920

/chat/completions Request URL

POST
https://genaiapi.cloudsway.net/v1/ai/{endpoint}/chat/completions

Request Header

Parameter	Value
Authorization	Bearer
Content-Type	application/json

Request Body

Request Parameters

model string Required

Model name.

messages array Required

Context passed to the model, arranged in conversation order.

Message Types

System Message object (Optional)

System message, used to set the model's role, tone, task objectives, or constraints. Typically placed as the first element in the messages array.

It is not recommended to set System Message for QwQ models. System Message will not take effect for QVQ models.

Properties

role string Required

Role of the system message, fixed as system.

content string Required

System instructions, used to define the model's role, behavior guidelines, response style, and task constraints.

User Message object (Required)

User message, used to pass questions, instructions, or context to the model.

Properties

role string Required

Role of the user message, fixed as user.

content string / array Required

Message content. If the input contains only text, it is of string type; if the input contains multimodal data such as images, or explicit caching is enabled, it is of array type.

When using multimodal models or enabling explicit caching, content is an array where each element contains:

type string Required

Possible values: - text: Set to text when inputting text - image_url: Set to image_url when inputting images - input_audio: Set to input_audio when inputting audio - video: Set to video when inputting video as a list of images - video_url: Set to video_url when inputting video files

text string

Input text. Required when type is text.

image_url object

Input image information. Required when type is image_url.

Properties

url string Required

Image URL or Base64 Data URL.

input_audio object

Input audio information. Required when type is input_audio.

Properties

data string Required

Audio URL or Base64 Data URL.

format string Required

Format of the input audio, such as mp3, wav, etc.

video array

Input video information in the form of an image list. Required when type is video.

video_url object

Input video file information. Required when type is video_url.

Properties

url string Required

Public URL or Base64 Data URL of the video file.

fps float (Optional)

Frames per second for extraction. Value range: [0.1, 10], default: 2.0.

min_pixels integer (Optional)

Sets the minimum pixel threshold for input images or video frames.

max_pixels integer (Optional)

Sets the maximum pixel threshold for input images or video frames.

total_pixels integer (Optional)

Limits the total pixels of all frames extracted from the video (single frame pixels × total number of frames).

cache_control object (Optional)

Used to enable explicit caching.

Properties

type string Required

Only supports being set to ephemeral.

Assistant Message object (Optional)

Model's response. Typically used as context passed back to the model in multi-turn conversations.

Properties

role string Required

Role of the assistant message, fixed as assistant.

content string (Optional)

Text content of the model's response. Can be empty when tool_calls is included; otherwise content is required.

partial boolean (Optional) Default: false

Whether to enable prefix continuation.

tool_calls array (Optional)

Tool and parameter information returned after initiating Function Calling.

Properties

id string Required

ID of the tool response.

type string Required

Tool type, currently only supports being set to function.

function object Required

Tool and parameter information.

Properties

name string Required

Tool name.

arguments string Required

Parameter information, as a JSON format string.

index integer Required

Index of the current tool information in the tool_calls array.

Tool Message object (Optional)

Output information from the tool.

Properties

role string Required

Fixed as tool.

content string Required

Output content of the tool function, must be a string.

tool_call_id string Required

The id returned after initiating Function Calling, used to identify which tool the Tool Message corresponds to.

stream boolean Optional Default: false

Whether to reply in streaming output mode.

Possible values: - false: Returns all content at once after the model finishes generating - true: Outputs while generating, returning a data chunk for each portion of content generated

Recommended to set to true to improve reading experience and reduce timeout risk.

Note

If a non-streaming call does not complete within 300 seconds, the service will interrupt the request and return the content generated so far (rather than returning an error). It is strongly recommended to use streaming calls for scenarios with longer outputs.

stream_options object Optional

Configuration options for streaming output, only effective when stream is true.

Properties

include_usage boolean (Optional) Default: false

Whether to include token consumption information in the last data chunk of the response.

During streaming output, token consumption information can only appear in the last data chunk of the response.

temperature float Optional

Sampling temperature, controls the diversity of text generated by the model. Higher temperature produces more diverse text; lower temperature produces more deterministic text.

Value range: [0, 2)

Both temperature and top_p can control the diversity of generated text. It is recommended to set only one of them.

temperature Default Values

Qwen3.7 (non-thinking mode), Qwen3.6 (non-thinking mode), Qwen3.5 (non-thinking mode), Qwen3 (non-thinking mode), qwen-max series, qwen-plus series (non-thinking mode), qwen-flash series (non-thinking mode), qwen-turbo series (non-thinking mode), qwen open-source series, qwen-coder series, qwen-doc-turbo, Qwen3-VL (non-thinking): 0.7
QVQ series: 0.5
qwen-vl series, qwen2.5-omni-7b: 0.01
qwen-math series: 0
Qwen3.7 (thinking mode), Qwen3.6 (thinking mode), Qwen3.5 (thinking mode), Qwen3 (thinking mode), QwQ series: 0.6
qwen3-max-preview (thinking mode), qwen-long series: 1.0

top_p float Optional

Nucleus sampling probability threshold, controls the diversity of text generated by the model. Higher top_p produces more diverse text; lower top_p produces more deterministic text.

Value range: (0, 1.0]

Both temperature and top_p can control the diversity of generated text. It is recommended to set only one of them.

top_k integer Optional

Specifies the number of candidate tokens used for sampling during generation. Larger values produce more random output; smaller values produce more deterministic output. If set to null or greater than 100, the top_k strategy is disabled.

This parameter is not an OpenAI standard parameter. When calling via Python SDK, place it in the extra_body object. Configuration: extra_body={"top_k":xxx}.

repetition_penalty float Optional

Repetition degree in consecutive sequences during model generation. Increasing repetition_penalty reduces repetition in model output. 1.0 means no penalty. Must be greater than 0.

This parameter is not an OpenAI standard parameter. When calling via Python SDK, place it in the extra_body object. Configuration: extra_body={"repetition_penalty":xxx}.

presence_penalty float Optional

Controls content repetition when the model generates text.

Value range: [-2.0, 2.0]. Positive values reduce repetition, negative values increase repetition.

response_format object Optional Default: {"type": "text"}

Format of the returned content.

Possible values: - {"type": "text"}: Output text response - {"type": "json_object"}: Output standard format JSON string

If set to {"type": "json_object"}, you must explicitly instruct the model to output JSON in the prompt, e.g., "Please output in JSON format", otherwise an error will occur.

Properties

type string Required

Format of the returned content. Possible values: text / json_object.

max_tokens integer Optional Deprecated Soon

This parameter will be deprecated soon. For new integrations, please use max_completion_tokens.

Maximum length of the model's response (excluding chain-of-thought content). When the model's response exceeds this value, generation will stop early and the returned finish_reason will be length.

max_completion_tokens integer Optional

Maximum length of the model's output, including chain-of-thought and model response. When the model's output exceeds this value, generation will stop early and the returned finish_reason will be length.

Difference from max_tokens: max_completion_tokens limits the model's complete output (chain-of-thought + response), while max_tokens only limits the response portion. Thinking models are recommended to use max_completion_tokens.

vl_high_resolution_images boolean Optional Default: false

Whether to increase the pixel upper limit of input images to the pixel value corresponding to 16384 tokens.

true: Uses fixed resolution strategy, ignores max_pixels setting
false: Pixel upper limit is determined by max_pixels

This parameter is not an OpenAI standard parameter. When calling via Python SDK, place it in the extra_body object. Configuration: extra_body={"vl_high_resolution_images":xxx}.

n integer Optional Default: 1

Number of responses to generate, value range is 1-4. Suitable for scenarios requiring multiple candidate responses.

Only supported for Qwen3 (non-thinking mode) and qwen-plus-character models. If the tools parameter is passed, set n to 1.

thinking object Optional Default: {"type":"adaptive"}

Properties

thinking.type string

Possible values: - adaptive: Adaptive (default), the model autonomously determines whether thinking is needed - disabled: Disable thinking, respond directly

This parameter is not an OpenAI standard parameter. When calling via Python SDK, place it in the extra_body object.

preserve_thinking boolean Optional Default: false

Whether to concatenate the reasoning_content from assistant messages in conversation history to the model input.

Important: When using qwen3.8-max, preserve_thinking defaults to true, and all reasoning_content from historical conversations must be passed back completely. This parameter is not an OpenAI standard parameter. When calling via Python SDK, place it in the extra_body object.

thinking_budget integer Optional

Maximum number of tokens for the thinking process. Applicable to Qwen3.7 and Qwen3.5.

This parameter is not an OpenAI standard parameter. When calling via Python SDK, place it in the extra_body object.

reasoning_effort string Optional

Controls the model's reasoning intensity. Different models support different possible values and defaults.

DeepSeek-V4, GLM series, and kimi/kimi-k3: high (default) / max
qwen3.8-max model: xhigh (default) / medium / low

This parameter is not an OpenAI standard parameter. When calling via Python SDK, place it in the extra_body object.

tool_stream boolean Optional Default: false

Only effective when stream=true. Controls whether complex tool parameters are output in streaming mode.

false: Complex tool parameters are output all at once (default behavior), complex formats are more accurate
true: Complex tool parameters are output in streaming mode, no timeout risk for complex formats

This parameter is not an OpenAI standard parameter. When calling via Python SDK, place it in the extra_body object.

seed integer Optional

Random number seed. Used to ensure reproducible results with the same input and parameters.

Value range: [0, 2³¹−1]

logprobs boolean Optional Default: false

Whether to return log probabilities of output tokens.

Content generated during the thinking phase (reasoning_content) will not return log probabilities.

top_logprobs integer Optional Default: 0

Specifies the number of highest-probability candidate tokens to return at each generation step. Value range: [0, 5]. Only effective when logprobs is true.

stop string / array Optional

Used to specify stop words. When the string or token_id specified by stop appears in the model-generated text, generation will terminate immediately.

tools array Optional

An array containing one or more tool objects for the model to call during Function Calling. When tools are set and the model determines a tool call is needed, the response will return tool information via tool_calls.

Function Calling

type string Required

Tool type, currently only supports being set to function.

function object Required

Properties

name string Required

Tool name. Only allows letters, numbers, underscores (_), and hyphens (-), maximum 64 tokens.

description string Required

Tool description information, helps the model determine when and how to call the tool.

parameters object (Optional) Default: {}

Parameter description for the tool, must be a valid JSON Schema. If the parameters field is empty, it means the tool has no input parameters.

tool_choice string / object Optional Default: auto

Tool selection strategy.

Possible values: - auto: The model autonomously selects the tool strategy - none: No tool calls desired - {"type": "function", "function": {"name": "the_function_to_call"}}: Force calling a specific tool

Models in thinking mode do not support forcing a specific tool call.

parallel_tool_calls boolean Optional Default: false

Whether to enable parallel tool calls.

clear_thinking boolean Optional Default: false

Controls whether the reasoning_content (thinking process) from historical turns is included as context input to the model in multi-turn conversations. Only supported by GLM series.

true: Ignore reasoning_content from historical turns
false (default): Preserve reasoning_content from historical turns

This parameter is not an OpenAI standard parameter. When calling via Python SDK, place it in the extra_body object.

Request Examples

Text Input (Non-streaming)

curl -X POST https://genaiapi.cloudsway.net/v1/ai/{endpoint}/chat/completions \
  -H "Authorization: Bearer ${your_ak}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.8-max",
    "messages": [
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Who are you?"}
    ]
}'

Streaming Output

curl -X POST https://genaiapi.cloudsway.net/v1/ai/{endpoint}/chat/completions \
  -H "Authorization: Bearer ${your_ak}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.8-max",
    "messages": [
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "Who are you?"}
    ],
    "stream": true,
    "stream_options": {
        "include_usage": true
    }
}'

Image Input

curl -X POST https://genaiapi.cloudsway.net/v1/ai/{endpoint}/chat/completions \
  -H "Authorization: Bearer ${your_ak}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen-vl-plus",
    "messages": [{
        "role": "user",
        "content": [
            {"type": "image_url", "image_url": {"url": "https://dashscope.oss-cn-beijing.aliyuncs.com/images/dog_and_girl.jpeg"}},
            {"type": "text", "text": "What is this"}
        ]
    }]
}'

Video Input (Image List)

curl -X POST https://genaiapi.cloudsway.net/v1/ai/{endpoint}/chat/completions \
  -H "Authorization: Bearer ${your_ak}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen-vl-max",
    "messages": [{
        "role": "user",
        "content": [
            {
                "type": "video",
                "video": [
                    "https://img.alicdn.com/imgextra/i3/O1CN01K3SgGo1eqmlUgeE9b_!!6000000003923-0-tps-3840-2160.jpg",
                    "https://img.alicdn.com/imgextra/i4/O1CN01BjZvwg1Y23CF5qIRB_!!6000000003000-0-tps-3840-2160.jpg",
                    "https://img.alicdn.com/imgextra/i4/O1CN01Ib0clU27vTgBdbVLQ_!!6000000007859-0-tps-3840-2160.jpg"
                ]
            },
            {"type": "text", "text": "Describe the specific process of this video"}
        ]
    }]
}'

Tool Calling (Function Calling)

First Request

curl -X POST https://genaiapi.cloudsway.net/v1/ai/{endpoint}/chat/completions \
  -H "Authorization: Bearer ${your_ak}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.8-max",
    "messages": [
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "What is the weather like in Hangzhou"}
    ],
    "tools": [
        {
            "type": "function",
            "function": {
                "name": "get_current_weather",
                "description": "Useful when you want to query the weather of a specified city.",
                "parameters": {
                    "type": "object",
                    "properties": {
                        "location": {
                            "type": "string",
                            "description": "City or district, such as Beijing, Hangzhou, Yuhang, etc."
                        }
                    },
                    "required": ["location"]
                }
            }
        }
    ]
}'

Second Request (Passing Tool Results)

curl -X POST https://genaiapi.cloudsway.net/v1/ai/{endpoint}/chat/completions \
  -H "Authorization: Bearer ${your_ak}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.8-max",
    "messages": [
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "What is the weather like in Hangzhou"},
        {
            "role": "assistant",
            "tool_calls": [
                {
                    "id": "call_xxx",
                    "type": "function",
                    "function": {
                        "name": "get_current_weather",
                        "arguments": "{\"location\":\"Hangzhou\"}"
                    },
                    "index": 0
                }
            ]
        },
        {
            "role": "tool",
            "tool_call_id": "call_xxx",
            "content": "{\"location\":\"Hangzhou\",\"temperature\":\"25℃\",\"condition\":\"Sunny\"}"
        }
    ],
    "tools": [
        {
            "type": "function",
            "function": {
                "name": "get_current_weather",
                "description": "Useful when you want to query the weather of a specified city.",
                "parameters": {
                    "type": "object",
                    "properties": {
                        "location": {
                            "type": "string",
                            "description": "City or district."
                        }
                    },
                    "required": ["location"]
                }
            }
        }
    ]
}'

Deep Thinking

curl -X POST https://genaiapi.cloudsway.net/v1/ai/{endpoint}/chat/completions \
  -H "Authorization: Bearer ${your_ak}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.8-max",
    "messages": [
        {"role": "user", "content": "Write a quicksort algorithm in Python"}
    ],
    "enable_thinking": true
}'

Structured Output (JSON)

curl -X POST https://genaiapi.cloudsway.net/v1/ai/{endpoint}/chat/completions \
  -H "Authorization: Bearer ${your_ak}" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.8-max",
    "messages": [
        {"role": "user", "content": "Please output in JSON format: name, age, occupation"}
    ],
    "response_format": {"type": "json_object"}
}'

Response Examples

Non-streaming Response Example

{
    "id": "chatcmpl-6ada9ed2-7f33-9de2-8bb0-78bd4035025a",
    "object": "chat.completion",
    "created": 1735120033,
    "model": "qwen3.8-max",
    "choices": [
        {
            "index": 0,
            "message": {
                "role": "assistant",
                "content": "I am a large-scale language model developed by Alibaba Cloud, my name is Qwen.",
                "reasoning_content": null,
                "tool_calls": null
            },
            "finish_reason": "stop",
            "logprobs": null
        }
    ],
    "usage": {
        "prompt_tokens": 3019,
        "completion_tokens": 104,
        "total_tokens": 3123,
        "prompt_tokens_details": {
            "cached_tokens": 2048
        }
    },
    "system_fingerprint": null
}

Streaming Response Chunk Example

{"id":"chatcmpl-e30f5ae7-3063-93c4-90fe-beb5f900bd57","choices":[{"delta":{"content":"","role":"assistant","tool_calls":null},"finish_reason":null,"index":0,"logprobs":null}],"created":1735113344,"model":"qwen3.8-max","object":"chat.completion.chunk","usage":null}
{"id":"chatcmpl-e30f5ae7-3063-93c4-90fe-beb5f900bd57","choices":[{"delta":{"content":"I am","role":null,"tool_calls":null},"finish_reason":null,"index":0,"logprobs":null}],"created":1735113344,"model":"qwen3.8-max","object":"chat.completion.chunk","usage":null}
{"id":"chatcmpl-e30f5ae7-3063-93c4-90fe-beb5f900bd57","choices":[{"delta":{"content":" a large-scale language model from Alibaba Cloud, my name is Qwen.","role":null,"tool_calls":null},"finish_reason":"stop","index":0,"logprobs":null}],"created":1735113344,"model":"qwen3.8-max","object":"chat.completion.chunk","usage":null}
{"id":"chatcmpl-e30f5ae7-3063-93c4-90fe-beb5f900bd57","choices":[],"created":1735113344,"model":"qwen3.8-max","object":"chat.completion.chunk","usage":{"completion_tokens":17,"prompt_tokens":22,"total_tokens":39,"prompt_tokens_details":{"cached_tokens":0}}}

/v1/messages Request URL

POST
https://genaiapi.cloudsway.net/{endpoint}/v1/messages

Parameter	Value
Authorization	Bearer
Content-Type	application/json

Request Body

Request Parameters

model string Required

Model ID to call.

max_tokens integer Required

Maximum number of tokens for model output.

Model Differences

qwen3.8-max / deepseek‑v4‑pro: max_tokens = response tokens + thinking tokens total; when thinking is enabled, max_tokens > thinking.budget_tokens

system string / array Optional

System prompt, sets the model's role and behavior.

Format Description

String form is equivalent to a single type=text content block;

When enabling explicit caching (cache_control), array format must be used.

[
    {
        "type":"text",
        "text":"You are an assistant",
        "cache_control":{"type":"ephemeral"}
    }
]

messages array Required

Conversation message array, complete session history.

Message Structure Description

role: user / assistant / system content: Supports strings, and also structured arrays (text, images, video, tool_use, tool_result)

Text block

{"type":"text","text":"xxx","cache_control":{"type":"ephemeral"}}

Image block (vision models)

{"type":"image","source":{"type":"url","url":"https://xxx.jpg"}}

Video block (vision models)

{"type":"video","source":{"type":"url","url":"https://xxx.mp4"}}

tool_use: Model returns tool call
tool_result: Tool execution result passed back to the model

stream boolean Optional Default: false

Whether to enable streaming output. - false: Non-streaming, returns complete response at once - true: SSE streaming incremental return, includes content_block_start, content_block_delta, content_block_stop, message_delta, message_stop, ping events.

temperature number Optional

Value range: [0,2), controls generation randomness.

top_p number Optional

Nucleus sampling probability threshold. It is recommended to set only one of temperature or top_p.

top_k integer Optional

Sampling candidate set size.

stop_sequences array Optional

Custom stop sequences, terminates generation early when this text is encountered.

Behavior Description

After hitting a stop sequence, stop_reason will still be end_turn, and the response will not return the matched stop text.

thinking object Optional

Deep thinking configuration.

Properties

thinking.type string Required - enabled: Enable deep thinking, returns thinking content block - disabled: Disable deep thinking

thinking.budget_tokens integer

Only effective when type=enabled, represents the maximum token budget for the thinking process, independent of max_tokens.

tools array Optional

Function Calling tool definition array.

Tool Structure Example

[
    {
        "name":"get_weather",
        "description":"Get city weather",
        "input_schema":{
            "type":"object",
            "properties":{"city":{"type":"string"}},
            "required":["city"]
        }
    }
]

tool_choice object Optional

Tool calling strategy.

Possible Values

{"type":"auto"}: Model autonomously decides whether to call tools (default)
{"type":"any"}: Force calling any tool
{"type":"none"}: Prohibit tool calls
{"type":"tool","name":"xxx"}: Force calling a specific tool

output_config object Optional

Platform extension field for reasoning intensity and structured output configuration.

Properties

effort string: high / max, controls reasoning intensity; only supported by some models.

format object

{
    "type":"json_schema",
    "schema":{
        "type":"object",
        "properties":{},
        "required":[],
        "additionalProperties":false
    }
}

Request Examples

Basic Non-streaming Call curl

curl -X POST "https://genaiapi.cloudsway.net/{endpoint}/v1/messages" \
-H "Content-Type: application/json" \
-H "Authorization: ${your_ak}" \
-d '{
    "model": "qwen3.8-max",
    "max_tokens": 1024,
    "system": "You are a helpful assistant",
    "messages": [{"role": "user", "content": "Who are you?"}],
    "thinking":{"type":"disabled"}
}'

Streaming Output curl

curl -X POST "https://genaiapi.cloudsway.net/{endpoint}/v1/messages" \
-H "Content-Type: application/json" \
-H "Authorization: ${your_ak}" \
--no-buffer \
-d '{
    "model": "qwen3.8-max",
    "max_tokens": 1024,
    "stream": true,
    "messages": [{"role": "user", "content": "Briefly introduce artificial intelligence"}],
    "thinking":{"type":"disabled"}
}'

Deep Thinking Enabled curl

curl -X POST "https://genaiapi.cloudsway.net/{endpoint}/v1/messages" \
-H "Content-Type: application/json" \
-H "Authorization: ${your_ak}" \
-d '{
    "model": "qwen3.8-max",
    "max_tokens":2048,
    "stream":true,
    "thinking":{"type":"enabled","budget_tokens":1024},
    "messages":[{"role":"user","content":"Analyze the development prospects of quantum computing"}]
}'

Image Understanding curl

curl -X POST "https://genaiapi.cloudsway.net/{endpoint}/v1/messages" \
-H "Content-Type: application/json" \
-H "Authorization: ${your_ak}" \
-d '{
    "model":"qwen3.8-max",
    "max_tokens":1024,
    "stream":true,
    "messages":[
        {
            "role":"user",
            "content":[
                {
                    "type":"image",
                    "source":{"type":"url","url":"https://xxx.jpg"}
                },
                {"type":"text","text":"Describe the image content"}
            ]
        }
    ],
    "thinking":{"type":"disabled"}
}'

Function Call curl

curl -X POST "https://genaiapi.cloudsway.net/{endpoint}/v1/messages" \
-H "Content-Type: application/json" \
-H "Authorization: ${your_ak}" \
-d '{
    "model":"qwen3.8-max",
    "max_tokens":1024,
    "tools":[
        {
            "name":"get_weather",
            "description":"Get city weather",
            "input_schema":{"type":"object","properties":{"city":{"type":"string"}},"required":["city"]}
        }
    ],
    "messages":[{"role":"user","content":"What is the weather like in Hangzhou today"}]
}'

Explicit Caching curl

curl -X POST "https://genaiapi.cloudsway.net/{endpoint}/v1/messages" \
-H "Content-Type: application/json" \
-H "Authorization: ${your_ak}" \
-d '{
    "model":"qwen3.8-max",
    "max_tokens":1024,
    "system":[
        {
            "type":"text",
            "text":"<≥1024 token long text>",
            "cache_control":{"type":"ephemeral"}
        }
    ],
    "messages":[{"role":"user","content":"What is the content of this code"}]
}'

JSON Schema Structured Output curl

curl -X POST "https://genaiapi.cloudsway.net/{endpoint}/v1/messages" \
-H "Content-Type: application/json" \
-H "Authorization: ${your_ak}" \
-d '{
    "model":"deepseek-v4-pro",
    "max_tokens":1024,
    "messages":[{"role":"user","content":"Extract email information: Zhang San zhangsan@example.com, wants to schedule a product demo next Tuesday at 2 PM"}],
    "output_config":{
        "format":{
            "type":"json_schema",
            "schema":{
                "type":"object",
                "properties":{
                    "name":{"type":"string"},
                    "email":{"type":"string"},
                    "demo_requested":{"type":"boolean"}
                },
                "required":["name","email","demo_requested"],
                "additionalProperties":false
            }
        }
    }
}'

Response Examples

Non-streaming Response

{
    "id": "msg_e2898f19-fc0e-4cb3-bd9b-5b7dc4ea3bc9",
    "type": "message",
    "role": "assistant",
    "model": "qwen3.8-max",
    "content": [
        {
            "type": "thinking",
            "thinking": "Let me analyze this question...",
            "signature": ""
        },
        {
            "type": "text",
            "text": "Hello! I am Tongyi Qwen..."
        }
    ],
    "stop_reason": "end_turn",
    "stop_sequence": null,
    "usage": {
        "input_tokens": 22,
        "output_tokens": 223,
        "cache_creation_input_tokens": 0,
        "cache_read_input_tokens": 0
    }
}

Response Field Description

content[]: Array, supports text text, thinking thinking process, tool_use tool call blocks
stop_reason: end_turn normal completion / max_tokens reached output limit / tool_use triggered tool call
usage: cache_creation_input_tokens tokens consumed for cache creation; cache_read_input_tokens tokens from cache hits.

Streaming SSE Event Fragment

{"type":"message_start","message":{"id":"msg_xxx","type":"message","role":"assistant","model":"qwen3.8-max","content":[],"usage":{"input_tokens":15,"output_tokens":0}}}
{"type":"content_block_start","index":0,"content_block":{"type":"thinking","thinking":"","signature":""}}
{"type":"content_block_delta","index":0,"delta":{"type":"thinking_delta","thinking":"Thinking content..."}}
{"type":"content_block_stop","index":0}
{"type":"content_block_start","index":1,"content_block":{"type":"text","text":""}}
{"type":"content_block_delta","index":1,"delta":{"type":"text_delta","text":"Model response text"}}
{"type":"content_block_stop","index":1}
{"type":"message_delta","delta":{"stop_reason":"end_turn","stop_sequence":null},"usage":{"input_tokens":15,"output_tokens":1078,"cache_creation_input_tokens":0,"cache_read_input_tokens":0}}
{"type":"message_stop"}

Streaming Event Description

message_start: Stream starts, only returns input_tokens
content_block_start: Single content block starts
content_block_delta: Incremental data; text_delta/thinking_delta/input_json_delta
content_block_stop: Single content block ends
message_delta: Complete usage statistics, stop_reason
message_stop: Stream ends
{"type":"ping"}: Keep-alive heartbeat, client should ignore directly.