MaaS_Cl_Sonnet_5_20240620

Model Overview

item	value
Model ID (model)	claude-sonnet-5
Status	Active (Latest); Released on 2026-06-30, to be retired no earlier than 2027-06-30
context window	1M tokens (default value, i. e. the maximum, with no smaller context variants available)
Maximum Input	The source document does not provide separate clarification (the 1M tokens mentioned in the context refers to the aggregate of input and output tokens)
Maximum Output	128K tokens (up to 300K under the Message Batches API with the output-300k-2026-03-24 beta)
input modality	Text, Image
output modality	Text
thinking pattern	Adaptive (enabled by default, the model will still perform reasoning even if the request does not carry the thinkingfield)
default effort	high
Knowledge Cutoff	Reliable knowledge as of January 2026; training data as of January 2026
Tokenizer	New tokenizer: The number of tokens in the same segment of text is approximately 30% higher than that of Sonnet 4.6

Request Protocol

https

Header

Parameter Name	Type	Required	Description
Content-Type	string	Yes	application/json
Authorization	string	Yes	Bearer ${your_AK}

Anthropic Messages

Request URL

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

Header (Additional to This Agreement)

Parameter Name	Type	Required	Description
anthropic-version	string	Yes	The API version number required by the origin server, which is currently 2023-06-01

Request Body Parameters

Parameter Name	Secondary Parameters	Three-level parameters	Type	Required	Description
model	-	-	string	Yes	Fixed as MaaS_Cl_Sonnet_5_20240620 (Original vendor model: claude-sonnet-5)
max_tokens	-	-	number	Yes	Output hard cap, including the total of thinking and main text (this model has thinking enabled by default, please adjust this value upward as needed); the model may stop naturally in advance; set it to 0to warm up the prompt cache without generating a response; minimum value: 0
messages	-	-	array of MessageParam	Yes	Input messages: the maximum number per single request is 100,000; consecutive messages from the same role will be merged
	role	-	string	Yes	user / assistant / system
	content	-	string or array of ContentBlockParam	Yes	The string is[{"type":"text", "text":". .."}]for short.
		type	string	Yes	text / image / document / search_result / thinking / redacted_thinking / tool_use / tool_result / server_tool_use / web_search_tool_result / web_fetch_tool_result
		text	string	Conditions	type= textText content when (minLength 1)
		source	object	Conditions	Source for type=image/document: base64 (data + media_type), url, or file_id; image media_type supports image/jpeg/image/png/image/gif/image/webp (Sonnet 5 row)
		id / name / input	string / string / map	Conditions	type= tool_use: Tool call ID, tool name (1–200 characters), and input parameters
		tool_use_id / content	string / string or array	Conditions	type= tool_result: corresponds to the id of tool_use and the postback content (Sonnet 5 row)
		signature / thinking	string / string	Conditions	type= thinking: The signature and thinking text returned in the previous round must be posted back as-is; any modification will result in a 400 response invalid_request_error
		data	string	Conditions	type= redacted_thinkingencrypted content, postback as-is
		cache_control	object	No	Create a cache breakpoint at the content block:{"type":"ephemeral", "ttl":"5m"\|"1h"}, the default ttl is 5m
		citations	array	No	Text citation blocks (char_location / page_location / content_block_location / web_search_result_location / search_result_location) (Sonnet 5 row)
system	-	-	string or array of TextBlockParam	No	System prompt; the Messages API does not have systemrole messages, and system prompts are processed via this field (Sonnet 5 row)
stream	-	-	boolean	No	Whether to return incrementally via SSE
stop_sequences	-	-	array of string	No	Custom stop sequence; when hit, stop_reasonwill be stop_sequence, and the hit sequence will be returned in stop_sequence
tools	-	-	array of ToolUnion	No	Tool Definition (Client tools: name, description, input_schema; server-side tools are also acceptable, such as web_search / web_fetch / code_execution / text_editor / json, as well as browser use and computer_toolset_20260801that are supported by this model but not by version 4.6)
	name	-	string	Yes	Tool Name (1–128 characters, ^[a-zA-Z0-9_-]{1,128}$)
	description	-	string	No	Tool description, please provide as detailed as possible
	input_schema	-	object	Yes	Tool input parameters JSON Schema (type: "object", properties, required)
	cache_control / strict / defer_loading / allowed_callers / input_examples / eager_input_streaming	-	Respective types	No	cache breakpoint, strong parameter validation, lazy loading, allowed callers, input parameter examples, input parameter streaming
tool_choice	-	-	object	No	{"type":"auto"} / {"type":"any"} / {"type":"tool","name":"..."} / {"type":"none"}; each branch can carry disable_parallel_tool_use (default false) (Sonnet 5 row)
thinking	-	-	object	No	The only valid values for this model are{"type":"adaptive"}(default behavior, can be accompanied by display: summarized| omitted) and{"type":"disabled"}(disable reasoning);{"type":"enabled", "budget_tokens": N}has been removed, and a 400 error will be returned if it is passed.
output_config	-	-	object	No	effort: low/ medium/ high (default, equivalent to omitting the parameter) / xhigh/ max;do not use adaptiveas the value for effort. format: the Structured Outputs format{"type":"json_schema", "schema":{. ..}}. This model does not support modifying effort mid-generation on a per-message basis (a beta feature exclusive to other model variants)
metadata	-	-	object	No	user_id: External user identifier (UUID/hash recommended, max. 512 characters; do not include name/email/phone number)
service_tier	-	-	string	No	autoor standard_only; note that this model does not support Priority Tier
container	-	-	object or string	No	Container reuse identifier; the object form can include idand skills (up to 20 entries: skill_id+ type: anthropic/ custom+ version) (Sonnet 5 row)
cache_control	-	-	object	No	Top-level cache control, automatically set a cache breakpoint for the last cacheable block
inference_geo	-	-	string	No	Region for inference processing; if left unspecified, the workspace default will be used default_inference_geo
temperature	-	-	number	No	This model returns 400 for non-default values; please delete this parameter and use system prompts to guide the style
top_p	-	-	number	No	This model returns 400 for non-default values; please delete this parameter
top_k	-	-	number	No	This model returns 400 for non-default values; please delete this parameter

Request Example

curl -X POST 'https://genaiapi.cloudsway.net/{endpoint}/v1/messages' \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer ${your_AK}' \
  -H 'anthropic-version: 2023-06-01' \
  -d '{
    "model": "MaaS_Cl_Sonnet_5_20240620",
    "max_tokens": 4096,
    "messages": [{"role": "user", "content": "Hello, tell me about yourself"}]
  }'

Control Thinking and Gears (Exclusive Terminology for This Model)

# Disable thinking (enabled by default)
  -d '{
    "model": "MaaS_Cl_Sonnet_5_20240620",
    "max_tokens": 4096,
    "thinking": {"type": "disabled"},
    "messages": [{"role": "user", "content": "Translate the following sentence into English: The weather is nice today"}]
  }'

# Increase thinking depth (effort replaces budget_tokens from the 4.6 era)
  -d '{
    "model": "MaaS_Cl_Sonnet_5_20240620",
    "max_tokens": 32768,
    "thinking": {"type": "adaptive"},
    "output_config": {"effort": "xhigh"},
    "messages": [{"role": "user", "content": "Design a distributed rate-limiting solution and provide a trade-off analysis"}]
  }'

Response Example

{
  "id": "msg_01AbCdEfGhIjKlMnOpQrStUv",
  "type": "message",
  "role": "assistant",
  "model": "MaaS_Cl_Sonnet_5_20240620",
  "content": [{"type": "text", "text": "Hello! I'm Claude, happy to help you."}],
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {"input_tokens": 16, "output_tokens": 28}
}

Example of Streaming Request

curl -X POST 'https://genaiapi.cloudsway.net/{endpoint}/v1/messages' \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer ${your_AK}' \
  -H 'anthropic-version: 2023-06-01' \
  -d '{
    "model": "MaaS_Cl_Sonnet_5_20240620",
    "max_tokens": 4096,
    "stream": true,
    "messages": [{"role": "user", "content": "Hello, tell me about yourself"}]
  }'

SDK Calling Method

from anthropic import Anthropic

client = Anthropic(
    api_key="${your_AK}",
    base_url="https://genaiapi.cloudsway.net/{endpoint}/v1",
)

message = client.messages.create(
    model="MaaS_Cl_Sonnet_5_20240620",
    max_tokens=4096,
    output_config={"effort": "medium"},
    messages=[{"role": "user", "content": "Hello, tell me about yourself"}],
)
print(message.content[0].text)

OpenAI Chat Completions

Origin server note: This protocol is provided by the official Anthropic OpenAI SDK compatibility layer, designed to "facilitate rapid testing and model capability comparison", and is not a long-term/production-grade solution; for full capabilities (PDF processing, citations, reasoning process, prompt cache), please use the native Messages protocol.

Request URL

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

Request Body Parameters

Parameter Name	Secondary Parameters	Type	Required	Description
model	-	string	Yes	Fixed as MaaS_Cl_Sonnet_5_2024062
messages	-	array	Yes	Dialogue messages;rolesupports system (or developer, which will be promoted and concatenated as the first system prompt), user, assistant, tool, function
	content	string or array	Yes	user supports string, text, image_url (url);image_url. detail, input_audio, fileare ignored. assistant supports string, text;refusalis ignored
	tool_calls	array	No	Tool invocation for assistant;function_callis also supported.
	tool_call_id	string	Conditions	role= tool/ functionis mandatory
max_tokens	-	integer	No	Fully supported (including both the thinking process and the main text in total)
max_completion_tokens	-	integer	No	Fully supported
stream	-	boolean	No	Fully supported
stream_options	-	object	No	Fully supported
stop	-	string or array	No	Non-blank stop-write sequence is valid
n	-	integer	No	Must be 1
tools	-	array	No	tools[n]. function's name/ description/ parametersare fully supported;strictis ignored (use native Structured Outputs for strict validation)
functions	-	array	No	Same as tools (OpenAI has deprecated this field, and it is recommended to use tools instead)
parallel_tool_calls	-	boolean	No	Fully supported
thinking	-	object	No	passed in via the extension field of the compatibility layer;this model only accepts{"type":"adaptive"}or{"type":"disabled"}, passing{"type":"enabled", "budget_tokens": N}will return 400; the compatibility layer does not return Claude's detailed reasoning process
temperature	-	number	No	The compatibility layer documentation describes this as 0–1, however this model returns 400 for non-default values, please remove this parameter
top_p	-	number	No	The compatibility layer documentation states full support, however this model returns 400 for non-default values, please remove this parameter
logprobs / top_logprobs	-	-	No	Ignored
metadata	-	object	No	Ignored
response_format	-	object	No	Ignored (use native Structured Outputs for JSON output)
prediction / presence_penalty / frequency_penalty / seed / service_tier / audio / logit_bias / store / user / modalities / reasoning_effort	-	-	No	are all ignored

Request Example

curl -X POST 'https://genaiapi.cloudsway.net/v1/ai/{endpointPath}/chat/completions' \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer ${your_AK}' \
  -d '{
    "model": "MaaS_Cl_Sonnet_5_20240620",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Hello, tell me about yourself"}
    ]
  }'

Response Example

{
  "id": "chatcmpl-abc123",
  "object": "chat.completion",
  "created": 1789000000,
  "model": "MaaS_Cl_Sonnet_5_20240620",
  "choices": [
    {
      "index": 0,
      "finish_reason": "stop",
      "message": {"role": "assistant", "content": "Hello! I'm Claude, happy to help you."}
    }
  ],
  "usage": {"prompt_tokens": 21, "completion_tokens": 19, "total_tokens": 40}
}

origin server description: compatibility layer choices[]always has a length of 1;usage. completion_tokens_details, usage. prompt_tokens_details, logprobs, service_tier, system_fingerprint, choices[]. message. refusal, choices[]. message. audioare always empty.

Example of Streaming Request

curl -X POST 'https://genaiapi.cloudsway.net/v1/ai/{endpointPath}/chat/completions' \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer ${your_AK}' \
  -d '{
    "model": "MaaS_Cl_Sonnet_5_20240620",
    "stream": true,
    "messages": [{"role": "user", "content": "Hello, tell me about yourself"}]
  }'

SDK Calling Method

from openai import OpenAI

client = OpenAI(
    api_key="${your_AK}",
    base_url="https://genaiapi.cloudsway.net/v1/ai/{endpointPath}",
)

response = client.chat.completions.create(
    model="MaaS_Cl_Sonnet_5_20240620",
    messages=[{"role": "user", "content": "Hello, tell me about yourself"}],
)
print(response.choices[0].message.content)

---

Three migration tips (origin server is clear, most likely to be encountered when customers access the service):

① Please stop including temperature/ top_p/ top_k; non-default values will trigger a 400 error;

② thinkingor adaptiveor disabled; the legacy combination of enabled+ budget_tokenswill trigger a 400 error, and the thinking depth has been replaced with output_config. effort;

③ The new tokenizer increases the token count for the same text by approximately 30%, so the max_tokensoriginally tuned based on 4.6 may cause truncation, and the token-based quota and cost need to be recalculated (note: security interception returns 200 with stop_reason: "refusal", not an error code).

MaaS_Cl_Sonnet_4.6_20240620

Model Overview

item	value
Model ID (model)	claude-sonnet-4-6
Status	Active (Legacy); released on 2026-02-17, to be retired no earlier than 2027-02-17
context window	1M tokens
Maximum Input	The source document does not provide separate clarification (the 1M tokens mentioned in the context refers to the aggregate of input and output tokens)
Maximum Output	128K tokens (up to 300K under the Message Batches API with the output-300k-2026-03-24 beta)
input modality	Text, Image
output modality	Text
thinking pattern	Adaptive (extended mode is deprecated)
default effort	high
Knowledge Cutoff	Reliable knowledge as of August 2025; training data as of January 2026

Request Protocol

https

Header

Parameter Name	Type	Required	Description
Content-Type	string	Yes	application/json
Authorization	string	Yes	Bearer ${your_AK}

Anthropic Messages

Request URL

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

Header (Additional to This Agreement)

Parameter Name	Type	Required	Description
anthropic-version	string	Yes	The API version number required by the origin server, which is currently 2023-06-01

Request Body Parameters

Parameter Name	Secondary Parameters	Three-level parameters	Type	Required	Description
model	-	-	string	Yes	Fixed as MaaS_Cl_Sonnet_4.6_20240620 (Original vendor model: claude-sonnet-4-6)
max_tokens	-	-	number	Yes	The maximum number of tokens to generate; the model may stop naturally before reaching this limit; set this to 0to warm up the prompt cache without generating a response; different models have different upper limits (minimum value: 0)
messages	-	-	array of MessageParam	Yes	Enter messages: the maximum number per single request is 100,000; consecutive messages from the same role will be merged; if the last message is from the assistant role, the response will be continued from it
	role	-	string	Yes	user / assistant / system
	content	-	string or array of ContentBlockParam	Yes	The string is[{"type":"text", "text":". .."}]for short. (Sonnet 4.6 row)
		type	string	Yes	Content block type: text / image / document / search_result / thinking / redacted_thinking / tool_use / tool_result / server_tool_use / web_search_tool_result / web_fetch_tool_result
		text	string	Conditions	type= textText content when (minLength 1)
		source	object	Conditions	Source for type=image/document: base64 (data + media_type), url, or file_id; image media_type supports image/jpeg/image/png/image/gif/image/webp
		id / name / input	string / string / map	Conditions	type= tool_use: tool call ID (^[a-zA-Z0-9_-]+$), tool name (1–200 characters), and input parameters
		tool_use_id / content	string / string or array	Conditions	type= tool_result: corresponds to the id of the corresponding tool_use and the postback content (which may contain blocks such as text, image, search_result, document, and tool_reference)
		signature / thinking	string / string	Conditions	type= thinking: The signature and thinking text returned in the previous round must be posted back as-is; any modification will result in a 400 invalid_request_error
		data	string	Conditions	type= redacted_thinkingencrypted content at the time, and perform postback as-is
		cache_control	object	No	Create a cache breakpoint for this content block: `{"type":"ephemeral", "ttl":"5m"
		citations	array	No	Text citation blocks (char_location / page_location / content_block_location / web_search_result_location / search_result_location)
system	-	-	string or array of TextBlockParam	No	System prompt; the Messages API does not have systemrole messages, so the system prompt uses this field
stream	-	-	boolean	No	Whether to return results incrementally via SSE (for details about the origin server, please refer to the streaming documentation)
stop_sequences	-	-	array of string	No	Custom stop sequence; when triggered, the stop_reasonwill be stop_sequence, and the matched sequence will be returned in the stop_sequencefield
tools	-	-	array of ToolUnion	No	Tool Definition (Client tools: name, description, input_schema; server-side tools such as versioned tool types like web_search / web_fetch / code_execution / text_editor / json are also supported)
	name	-	string	Yes	Tool Name (1–128 characters, ^[a-zA-Z0-9_-]{1,128}$)
	description	-	string	No	Tool description, please provide as detailed as possible
	input_schema	-	object	Yes	Tool input parameters JSON Schema (type: "object", properties, required)
	cache_control / strict / defer_loading / allowed_callers / input_examples / eager_input_streaming	-	Respective types	No	cache breakpoint, mandatory validation of input parameter schema, lazy loading, allowed callers, input parameter examples, input parameter streaming
tool_choice	-	-	object	No	Tool Selection:{"type":"auto"}(decided by the model itself) /{"type":"any"}(mandatory tool) /{"type":"tool", "name":". .."}(specified tool) /{"type":"none"}(tool disabled); each branch can carry disable_parallel_tool_use (default is false, when set to true, tool calls are limited to at most / exactly once)
thinking	-	-	object	No	Thinking configuration, choose one of three: {"type":"enabled","budget_tokens":N,"display":"summarized"\|"omitted"} (budget_tokens >=1024 and less than max_tokens), {"type":"disabled"}, {"type":"adaptive","display":...}; displaydefaults to summarized; when set to omitted, no reasoning content is returned, but the signature is retained to enable multi-turn continuation.
output_config	-	-	object	No	Output configuration: effort (low/medium/high/xhigh/max), format (Structured Outputs {"type":"json_schema","schema":{...}})
metadata	-	-	object	No	Request metadata: user_id (external user identifier, uuid/hash recommended, maximum length 512, do not include name/email/phone number)
service_tier	-	-	string	No	auto (use priority capacity when available) or standard_only
container	-	-	object or string	No	Container reuse identifier; the object form can include idand skills (up to 20 entries, each consisting of skill_id+ type: anthropic/ custom+ version)
cache_control	-	-	object	No	Top-level cache control, which automatically sets a cache breakpoint for the last cacheable block in a request
inference_geo	-	-	string	No	Region for inference processing; if left unspecified, the workspace default will be used default_inference_geo

Request Example

curl -X POST 'https://genaiapi.cloudsway.net/{endpoint}/v1/messages' \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer ${your_AK}' \
  -H 'anthropic-version: 2023-06-01' \
  -d '{
    "model": "MaaS_Cl_Sonnet_4.6_20240620",
    "max_tokens": 1024,
    "messages": [{"role": "user", "content": "Hello, tell me about yourself"}]
  }'

Response Example

{
  "id": "msg_01XyZabcdEFghijklMNOpqr",
  "type": "message",
  "role": "assistant",
  "model": "MaaS_Cl_Sonnet_4.6_20240620",
  "content": [{"type": "text", "text": "Hello! I'm Claude, an AI assistant trained by Anthropic."}],
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {"input_tokens": 18, "output_tokens": 32}
}

Example of Streaming Request

curl -X POST 'https://genaiapi.cloudsway.net/{endpoint}/v1/messages' \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer ${your_AK}' \
  -H 'anthropic-version: 2023-06-01' \
  -d '{
    "model": "MaaS_Cl_Sonnet_4.6_20240620",
    "max_tokens": 1024,
    "stream": true,
    "messages": [{"role": "user", "content": "Hello, tell me about yourself"}]
  }'

SDK Calling Method

from anthropic import Anthropic

client = Anthropic(
    api_key="${your_AK}",
    base_url="https://genaiapi.cloudsway.net/{endpoint}/v1",
)

message = client.messages.create(
    model="MaaS_Cl_Sonnet_4.6_20240620",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello, tell me about yourself"}],
)
print(message.content[0].text)

OpenAI Chat Completions

Origin server note: This protocol is provided by the official Anthropic OpenAI SDK compatibility layer, designed to "facilitate rapid testing and model capability comparison", and is not a long-term/production-grade solution; for full capabilities (PDF processing, citations, reasoning process, prompt cache), please use the native Messages protocol.

Request URL

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

Request Body Parameters

Parameter Name	Secondary Parameters	Type	Required	Description
model	-	string	Yes	Fixed as MaaS_Cl_Sonnet_4.6_20240620 (Original vendor model: claude-sonnet-4-6)
messages	-	array	Yes	Dialogue messages;rolesupports system (or developer, which will be promoted and concatenated as the first system prompt), user, assistant, tool, function
	content	string or array	Yes	user supports string, text, image_url (url);image_url. detail, input_audio, fileare ignored. assistant supports string, text;refusalis ignored
	tool_calls	array	No	Tool invocation for assistant;function_callis also supported.
	tool_call_id	string	Conditions	role= tool/ functionis mandatory
max_tokens	-	integer	No	Fully supported
max_completion_tokens	-	integer	No	Fully supported
stream	-	boolean	No	Fully supported
stream_options	-	object	No	Fully supported
temperature	-	number	No	The value ranges from 0 to 1 (inclusive); any value greater than 1 will be truncated to 1.
top_p	-	number	No	Fully supported
n	-	integer	No	Must be 1
stop	-	string or array	No	Non-blank stop-write sequence is valid
tools	-	array	No	tools[n]. function's name/ description/ parametersare fully supported;strictis ignored (use native Structured Outputs if strong validation is required)
functions	-	array	No	Same as tools (OpenAI has deprecated this field, and it is recommended to use tools instead)
parallel_tool_calls	-	boolean	No	Fully supported
thinking	-	object	No	Passed in via the extension field of the compatibility layer:{"type":"enabled", "budget_tokens": 2000}; the compatibility layer does not return Claude's detailed reasoning process
logprobs / top_logprobs	-	-	No	Ignored
metadata	-	object	No	Ignored
response_format	-	object	No	Ignored (use native Structured Outputs for JSON output)
prediction / presence_penalty / frequency_penalty / seed / service_tier / audio / logit_bias / store / user / modalities / reasoning_effort	-	-	No	are all ignored

Request Example

curl -X POST 'https://genaiapi.cloudsway.net/v1/ai/{endpointPath}/chat/completions' \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer ${your_AK}' \
  -d '{
    "model": "MaaS_Cl_Sonnet_4.6_20240620",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "Hello, tell me about yourself"}
    ]
  }'

Response Example

{
  "id": "chatcmpl-abc123",
  "object": "chat.completion",
  "created": 1789000000,
  "model": "MaaS_Cl_Sonnet_4.6_20240620",
  "choices": [
    {
      "index": 0,
      "finish_reason": "stop",
      "message": {"role": "assistant", "content": "Hello! I'm Claude, happy to help you."}
    }
  ],
  "usage": {"prompt_tokens": 20, "completion_tokens": 18, "total_tokens": 38}
}

origin server description: compatibility layer choices[]always has a length of 1;usage. completion_tokens_details, usage. prompt_tokens_details, logprobs, service_tier, system_fingerprint, choices[]. message. refusal, choices[]. message. audioare always empty.

Example of Streaming Request

curl -X POST 'https://genaiapi.cloudsway.net/v1/ai/{endpointPath}/chat/completions' \
  -H 'Content-Type: application/json' \
  -H 'Authorization: Bearer ${your_AK}' \
  -d '{
    "model": "MaaS_Cl_Sonnet_4.6_20240620",
    "stream": true,
    "messages": [{"role": "user", "content": "Hello, tell me about yourself"}]
  }'

SDK Calling Method

from openai import OpenAI

client = OpenAI(
    api_key="${your_AK}",
    base_url="https://genaiapi.cloudsway.net/v1/ai/{endpointPath}",
)

response = client.chat.completions.create(
    model="MaaS_Cl_Sonnet_4.6_20240620",
    messages=[{"role": "user", "content": "Hello, tell me about yourself"}],
)
print(response.choices[0].message.content)