Skip to main content
POST
/
v1
/
messages
curl --request POST \
  --url https://api.getinfinityblue.com/v1/messages \
  --header 'Authorization: Bearer <token>' \
  --header 'Content-Type: application/json' \
  --header 'anthropic-version: <anthropic-version>' \
  --data '
{
  "model": "claude-3-opus-20240229",
  "max_tokens": 1024,
  "messages": [
    {
      "role": "user",
      "content": "Introduce yourself in one sentence."
    }
  ]
}
'
{
  "id": "<string>",
  "type": "<string>",
  "role": "<string>",
  "content": [
    {
      "type": "<string>",
      "text": "<string>"
    }
  ],
  "model": "<string>",
  "usage": {
    "input_tokens": 123,
    "output_tokens": 123,
    "cache_creation_input_tokens": 123,
    "cache_read_input_tokens": 123
  }
}

​
Models you can use

Model IDNotes
claude-3-opus-20240229Claude 3 flagship — top reasoning and comprehension
claude-3-5-sonnet-20241022Claude 3.5 Sonnet — balanced performance
claude-3-5-haiku-20241022Claude 3.5 Haiku — fast and lightweight
See GET /v1/models for the full list.

​
Version header

Every request must include anthropic-version: 2023-06-01 (or a newer version).

​
System prompt

The system field can be a plain text string or an array of content blocks with cache-control directives for fine-grained prompt caching.

​
Tool calling

Define tools in tools using a JSON Schema input_schema. The model returns tool_use blocks in content; feed back the tool result as a tool_result block in the next turn.

​
Extended thinking

Set thinking.type: enabled and specify thinking.budget_tokens to enable extended thinking mode. The model returns thinking content blocks alongside the final answer.

​
Streaming

Set stream: true for SSE streaming. Event types include message_start, content_block_delta, and message_stop.

Authorizations

​
Authorization
string
header
required

Bearer token authentication, format: Authorization: Bearer sk-xxxxxx. Get your API key in the console.

Headers

​
anthropic-version
string
required

Anthropic API version. Recommended value is 2023-06-01.

Example:

"2023-06-01"

​
x-api-key
string

Anthropic API key (optional). You can use this header or the standard Authorization: Bearer YOUR_KEY header — either works.

Body

application/json

Anthropic Claude Messages API request body.

​
model
string
required

Model ID, e.g. claude-3-opus-20240229. See GET /v1/models for the full list.

Example:

"claude-3-opus-20240229"

​
messages
object[]
required

Conversation messages. role must alternate between user and assistant.

​
max_tokens
integer
required

Maximum number of tokens the model may generate. Required.

Required range: x >= 1
​
system

System prompt — either a plain text string or an array of content blocks with cache-control directives.

​
temperature
number

Sampling temperature between 0 and 1, controlling output randomness.

Required range: 0 <= x <= 1
​
top_p
number

Nucleus sampling threshold. Tune this or temperature, not both.

​
top_k
integer

Top-K sampling — only the K most probable tokens are considered.

​
stream
boolean

Whether to stream the response as Server-Sent Events.

​
stop_sequences
string[]

Stop sequences — generation halts when any of these is produced.

​
tools
object[]

Tools the model may call. Each tool requires name, description, and input_schema.

​
tool_choice
object

Tool calling strategy. auto lets the model decide, any forces at least one tool call, tool forces a specific tool (also set name).

​
thinking
object

Extended thinking configuration, only effective for models that support it.

​
metadata
object

Additional request metadata.

Response

Successful response

Anthropic Claude Messages API response body.

​
id
string

Unique identifier for this message.

​
type
string

Object type, value is message.

Example:

"message"

​
role
string

Response role, value is assistant.

Example:

"assistant"

​
content
object[]

List of response content blocks — may include text, tool_use, or thinking blocks.

​
model
string

The model that actually processed the request.

​
stop_reason
enum<string>

The reason generation stopped.

Available options:
end_turn,
max_tokens,
stop_sequence,
tool_use
​
usage
object

Token usage statistics.