Anthropic Messages API

POST

/apps/anthropic/v1/messages

import anthropic
import os

client = anthropic.Anthropic(
    api_key=os.getenv("DASHSCOPE_API_KEY"),
    base_url="https://dashscope-intl.aliyuncs.com/apps/anthropic",
)

message = client.messages.create(
    model="qwen3.8-max",
    max_tokens=1024,
    system="You are a helpful assistant",
    messages=[
        {
            "role": "user",
            "content": "Who are you?"
        }
    ],
    thinking={"type": "disabled"},
)

print(message.content[0].text)

{
  "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 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
  }
}

FAQ

After configuring Claude Desktop or Claude Code, the connection test fails with Model discovery — Gateway /v1/models returned HTTP 404, or the request URL contains /v1/v1/models. How do I fix it? The model discovery feature of clients such as Claude Desktop and Claude Code automatically appends /v1/models to the configured base URL. Check the following two points:

Do not end the base URL with /v1/: it should end at /apps/anthropic (for example, for Singapore use https://dashscope-intl.aliyuncs.com/apps/anthropic). If you mistakenly enter .../apps/anthropic/v1/, the client appends /v1/models and produces the duplicated path /v1/v1/models, which returns HTTP 404.
Add models manually to skip discovery: the Anthropic-compatible endpoint provides only the Messages API (/v1/messages) and does not provide a model list endpoint (/v1/models), so the model discovery request returns 404 as well. Manually add models (for example, qwen3.7-plus) under Models in the client to skip automatic discovery.

Authorizations

string

header

required

QwenCloud API key passed via x-api-key header. Authorization: Bearer header is also supported.

Body

application/json

string

required

Model name. Supported models:

Qwen Max: qwen3.8-max, qwen3.7-max, qwen3.7-max-2026-06-08, qwen3.7-max-2026-05-20, qwen3.6-max-preview, qwen3-max, qwen3-max-2026-01-23, qwen3-max-preview

Qwen Plus: qwen3.6-plus, qwen3.6-plus-2026-04-02, qwen3.5-plus, qwen3.5-plus-2026-04-20, qwen3.5-plus-2026-02-15, qwen-plus, qwen-plus-latest, qwen-plus-2025-09-11

Qwen Flash: qwen3.7-flash, qwen3.7-flash-2026-07-15, qwen3.6-flash, qwen3.6-flash-2026-04-16, qwen3.5-flash, qwen3.5-flash-2026-02-23, qwen-flash, qwen-flash-2025-07-28

Qwen Turbo: qwen-turbo, qwen-turbo-latest

Qwen Coder: qwen3-coder-next, qwen3-coder-plus, qwen3-coder-plus-2025-09-23, qwen3-coder-flash

Qwen VL: qwen3-vl-plus, qwen3-vl-flash, qwen-vl-max, qwen-vl-plus

Qwen Open-source: qwen3.6-27b, qwen3.5-397b-a17b, qwen3.5-122b-a10b, qwen3.5-27b, qwen3.5-35b-a3b, qwen3.8-2.4t-a95b, qwen3.8-27b

Third-party models: deepseek-v4-pro-0813, deepseek-v4-pro, deepseek-v4-flash, deepseek-v4-flash-0731, deepseek-v3.2

integer

required

deepseek-v4-pro-0813, deepseek-v4-pro, deepseek-v4-flash, deepseek-v4-flash-0731, qwen3.8-max: max_tokens is the maximum total token count for both the reply content and the chain-of-thought content. Generation stops early when the model output exceeds this value, and stop_reason is max_tokens. max_tokens limits the combined length of the reply content and the thinking process. When extended thinking is enabled, max_tokens must be greater than thinking.budget_tokens.

glm-5.2: When the thinking.budget_tokens parameter is not passed, max_tokens is the maximum total token count for both the reply content and the chain-of-thought content. Generation stops early when the model output exceeds this value, and stop_reason is max_tokens. When the thinking.budget_tokens parameter is passed, max_tokens is only the maximum number of tokens for the reply content, and the thinking tokens are controlled separately by thinking.budget_tokens.

Other models: Maximum number of tokens for the reply content. If the generated content exceeds this value, generation stops early and stop_reason is max_tokens. max_tokens does not limit the length of the thinking process. When extended thinking is enabled, the thinking tokens are controlled separately by thinking.budget_tokens.

object[]

required

The message array, arranged in conversational order.

Show child attributes

enum<string>

required

Message role.

Available options:user,assistant,system

string

required

Message content. Can be a plain text string or a structured content array.

string

System prompt to set the model's role or behavior. Passed as a top-level parameter; the messages array does not accept the system role. A string is equivalent to a single type="text" content block. To use context caching, pass an array of content blocks with cache_control.

boolean

defaultfalse

Enable streaming output. Default is false.

number

Controls diversity of generated text, range [0, 2). Higher values produce more random output. This range differs from Anthropic's native [0.0, 1.0] — verify this parameter when migrating from Anthropic.

qwen3.8-max: Defaults to 0.6 in thinking mode. Values below 0.6 are automatically adjusted to 0.6.

number

Probability threshold for nucleus sampling. Both temperature and top_p control diversity — set only one at a time.

integer

Size of the sampling candidate set during generation.

string[]

Text sequences that stop generation. The model stops before outputting the sequence. When hit, stop_reason is still end_turn and the matched sequence is not included in the response.

object

Extended thinking configuration. When enabled, the model reasons before responding, and the response includes thinking-type content blocks. Not all models support thinking mode.

Show child attributes

enum<string>

enabled (enable thinking) or disabled (disable thinking).

Available options:enabled,disabled

integer

deprecated

To be deprecated. For new integrations, use output_config.effort. Maximum tokens for the thinking process. Does not overlap with max_tokens: this parameter limits thinking; max_tokens limits the final reply. Takes effect when type is enabled.

object[]

Tool definition array for function calling.

Show child attributes

string

required

Tool name.

object

required

JSON Schema definition for the tool's input parameters.

string

Description of the tool's functionality.

object

Tool choice strategy. {"type": "auto"}: model decides whether to call tools (default). {"type": "any"}: force calling any tool. {"type": "none"}: disable tool calling. {"type": "tool", "name": "tool_name"}: force calling a specific tool.

Show child attributes

enum<string>

Strategy type.

Available options:auto,any,none,tool

string

When type is tool, specifies the name of the tool to call.

object

Show child attributes

enum<string>

Controls the inference intensity of models. The valid values and default values vary by model.

glm-5.2, deepseek-v4-pro-0813, deepseek-v4-pro, and deepseek-v4-flash (including deepseek-v4-flash-0731): Default value: max. Valid values: high and max. low and medium are mapped to high, and xhigh is mapped to max.
qwen3.8-max: Default value: xhigh. Valid values: xhigh, medium, and low. max and high are mapped to xhigh.

Available options:low,medium,high,xhigh,max

object

Structured output configuration. When enabled, the model returns a JSON string. Behavior varies by model:

Strict structured outputs: Available for qwen3.8, qwen3.7, deepseek, and glm series models. The model strictly follows the provided JSON Schema, guaranteeing the same field types and hierarchy.
Regular structured outputs: For all other models, schema field constraints are not enforced — the API automatically falls back to a plain JSON mode (only guaranteeing that the output is a valid JSON string). In this fallback mode, the request must satisfy both of the following: (1) the output_config parameter is explicitly provided; (2) the system or messages content contains the keyword "JSON" (case-insensitive). If the keyword "JSON" is missing, the API throws: 'messages' must contain the word 'json' in some form.

Show child attributes

enum<string>

required

Fixed value: json_schema.

Available options:json_schema

object

required

JSON Schema object that follows the standard JSON Schema specification. Should include type (data type), properties (field definitions), required (array of required field names), and additionalProperties (must be set to false).

Response

200-application/json

string

Unique message identifier.

enum<string>

Always message.

Available options:message

enum<string>

Always assistant.

Available options:assistant

string

The model used for generation.

object[]

Content array. Element types can be text, thinking (returned when thinking is enabled), or tool_use (tool call).

Show child attributes

enum<string>

Content block type.

Available options:text,thinking,tool_use

string

Model-generated text reply when type is text.

string

The model's reasoning before the final response.

string

Signature when type is thinking. Currently always an empty string.

string

Unique identifier for the tool call when type is tool_use.

string

Name of the tool being called when type is tool_use.

object

Tool call input parameters when type is tool_use.

enum<string>

Stop reason: end_turn (normal completion), max_tokens (token limit reached), tool_use (tool call).

Available options:end_turn,max_tokens,tool_use

string | null

Always null.

object

Token usage statistics. In streaming, the usage in the message_start event only contains input_tokens and output_tokens; all 4 fields appear in the message_delta event.

Show child attributes

integer

Number of input tokens.

integer

Number of output tokens.

integer

Number of input tokens consumed for cache creation.

integer

Number of input tokens consumed by cache reads.

​FAQ

Authorizations

Body

Response

FAQ