Messages

POST /v1/messages

Generate a reply using the Anthropic Messages protocol. Send a list of conversation messages and the model returns the next assistant message. Ideal when you're using the Anthropic SDK or Claude Code.

This endpoint is compatible with the Anthropic Messages SDK. Point base_url at CrossModel and set model to a CrossModel model ID. Auth uses the x-api-key header.

Endpoint

POST https://api.crossmodel.ai/v1/messages
x-api-key: cm-YOUR_KEY
anthropic-version: 2023-06-01
Content-Type: application/json

Each request needs an API key in the x-api-key header (Authorization: Bearer also works). Create and manage keys on the console's API Keys page.

Create message

1curl https://api.crossmodel.ai/v1/messages \2  -H "Content-Type: application/json" \3  -H "anthropic-version: 2023-06-01" \4  -H "x-api-key: $CROSSMODEL_API_KEY" \5  -d '{6    "model": "anthropic/claude-sonnet-4.6",7    "max_tokens": 1024,8    "system": "You are a helpful assistant.",9    "messages": [10      { "role": "user", "content": "Hello!" }11    ]12  }'

Request parameters

Parameter	Type	Required	Notes
`model`	string	Yes	The model ID to use, e.g. `anthropic/claude-sonnet-4.6`. Call `/v1/models` for the full list.
`messages`	array	Yes	The conversation so far, in chronological order, with at least one entry. See Message object below.
`max_tokens`	integer	Yes	Max tokens to generate for this reply.
`system`	string or array	No	A system prompt that sets the model's behavior and persona.
`temperature`	number	No	Sampling temperature, typically `0`–`1`. Higher is more random and varied.
`top_p`	number	No	Nucleus-sampling threshold. Usually adjust either `temperature` or `top_p`, not both.
`top_k`	integer	No	Sample only from the K highest-probability candidates.
`stop_sequences`	string[]	No	Custom stop sequences. Generation halts at the first match.
`tools`	array	No	The tools the model may call. See Function tools below.
`tool_choice`	object	No	Controls whether and how the model calls tools.
`metadata`	object	No	Request metadata; may include `user_id` for safety and abuse detection.
`thinking`	object	No	Extended-thinking config. `type` is `enabled`, `disabled`, or `adaptive`; when `enabled`, provide `budget_tokens`. The effect depends on the model.
`stream`	boolean	No	Whether to stream the result. Default `false`.

Message object

messages is a chronological array with at least one message.

Field	Type	Required	Notes
`role`	string	Yes	The sender's role; only `user` and `assistant`. Put the system prompt in the top-level `system` field, not in `messages`.
`content`	string or array	Yes	The message content. Use a string for plain text; use a content-block array for images or tool content (see Content blocks).

Content blocks

When content is an array, each element is a content block with a type field:

type	Fields	Notes
`text`	`text`	A piece of text.
`image`	`source`	Image input. `source` is `{ "type": "base64", "media_type": ..., "data": ... }` or `{ "type": "url", "url": ... }`.
`tool_use`	`id`, `name`, `input`	A tool call issued by the `assistant`.
`tool_result`	`tool_use_id`, `content`, `is_error`	The result of running a tool; `tool_use_id` points at the corresponding `tool_use`.

Document and PDF input (the document content block) isn't supported yet.

Function tools

Declare functions the model can call via the tools field. The model returns a tool_use content block when needed; your code runs it and passes the result back as a tool_result content block.

{
  "tools": [
    {
      "name": "get_weather",
      "description": "Get current weather for a city.",
      "input_schema": {
        "type": "object",
        "properties": {
          "city": { "type": "string" }
        },
        "required": ["city"]
      }
    }
  ]
}

Field	Type	Required	Notes
`name`	string	Yes	The function name the model uses to reference the tool.
`description`	string	No	What the function does. A clear description helps the model decide when to call it.
`input_schema`	object	No	JSON Schema for the function's parameters.

Example request

curl https://api.crossmodel.ai/v1/messages \
  -H "x-api-key: cm-YOUR_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic/claude-sonnet-4.6",
    "max_tokens": 1024,
    "system": "You are a helpful assistant.",
    "messages": [
      { "role": "user", "content": "Hello!" }
    ]
  }'

Response

A non-streaming request returns a message object.

{
  "id": "msg_abc123",
  "type": "message",
  "role": "assistant",
  "model": "anthropic/claude-sonnet-4.6",
  "content": [
    { "type": "text", "text": "Hello! How can I help?" }
  ],
  "stop_reason": "end_turn",
  "stop_sequence": null,
  "usage": {
    "input_tokens": 18,
    "output_tokens": 14,
    "cache_creation_input_tokens": 0,
    "cache_read_input_tokens": 0
  }
}

Field	Type	Notes
`id`	string	A unique ID for this reply.
`type`	string	Always `message`.
`role`	string	Always `assistant`.
`model`	string	The model ID used for this request.
`content`	array	The generated content blocks.
`stop_reason`	string	Why it stopped — see below.
`stop_sequence`	string or null	The stop sequence that was matched, or `null` if none.
`usage`	object	Token usage for this request, used for billing.

Response content blocks

In the response content array, each element is a content block with a type field:

type	Fields	Notes
`text`	`text`	The generated text.
`tool_use`	`id`, `name`, `input`	A function call issued by the model. `input` is the arguments object.

stop_reason

Value	Notes
`end_turn`	The model finished its reply naturally.
`max_tokens`	Hit the `max_tokens` cap.
`stop_sequence`	Matched a stop sequence.
`tool_use`	The model switched to issuing a tool call.

The usage object

Field	Type	Notes
`input_tokens`	integer	Input tokens.
`output_tokens`	integer	Output tokens.
`cache_creation_input_tokens`	integer	Input tokens written to the cache.
`cache_read_input_tokens`	integer	Input tokens that hit the cache.

Streaming response

Set stream to true and the endpoint returns text/event-stream. A reply is made up of these events, in order:

Event	Notes
`message_start`	The message begins, with initial message info.
`content_block_start`	A content block begins.
`content_block_delta`	An increment of a content block (`text_delta`, `input_json_delta`, or `thinking_delta`).
`content_block_stop`	A content block ends.
`message_delta`	Carries `stop_reason` and `usage`.
`message_stop`	The message ends.

curl https://api.crossmodel.ai/v1/messages \
  -H "x-api-key: cm-YOUR_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "anthropic/claude-sonnet-4.6",
    "max_tokens": 1024,
    "messages": [{ "role": "user", "content": "Hi" }],
    "stream": true
  }'

Errors

Errors use the Anthropic format:

{
  "type": "error",
  "error": {
    "type": "authentication_error",
    "message": "Missing API key."
  }
}

HTTP status	`error.type`	Notes
`400`	`invalid_request_error`	Malformed body JSON, message structure, or parameters.
`401`	`authentication_error`	API key missing or invalid.
`402`	`permission_error`	Out of balance or not permitted.
`404`	`invalid_request_error`	The requested model doesn't exist or is currently unavailable.
`429`	`rate_limit_error`	Too many requests — rate limited.
`502`	`api_error`	The model service returned an error or an abnormal response.
`503`	`overloaded_error`	Model temporarily unavailable — retry shortly.
`500`	`api_error`	A CrossModel internal error.