Messages API | Shark AI

Language Models

Anthropic-Compatible Messages Format

Anthropic-compatible messages for models that expose Claude-style content blocks, tool use, streaming events, and thinking blocks.

Base URL: https://shark.ai/api/v1

Endpoint

POST /messages

Use when

The selected model supports anthropic_messages, especially for Claude-style tool use, thinking blocks, and signature preservation.

Protocol Support

Only call this endpoint for models whose supportedProtocols include anthropic_messages. For generic OpenAI SDK compatibility, use OpenAI Format.

Request Body

Parameter	Type	Required	Description
model	string	yes	Model ID, for example anthropic/claude-opus-4.7
messages	array	yes	User and assistant messages. Content can be a string or content block array.
max_tokens	integer	yes	Maximum output tokens.
system	string \| array		System prompt or system content blocks.
stream	boolean		Enable Anthropic-style SSE streaming.
temperature	number		Sampling temperature.
top_p	number		Nucleus sampling.
top_k	integer		Top-k sampling where supported.
stop_sequences	array		Stop sequences.
tools	array		Tool definitions using Anthropic input_schema.
tool_choice	object		Tool selection policy.
thinking	object		Extended thinking configuration where supported.

Tool Use

Define tools with Anthropic's input_schema. The model returns tool_use blocks; send tool results back as tool_result blocks in a later user message.

{
  "model": "anthropic/claude-opus-4.7",
  "max_tokens": 1024,
  "messages": [
    {
      "role": "user",
      "content": "What is the weather in Tokyo?"
    }
  ],
  "tools": [
    {
      "name": "get_weather",
      "description": "Get current weather for a city",
      "input_schema": {
        "type": "object",
        "properties": {
          "city": {
            "type": "string"
          }
        },
        "required": [
          "city"
        ]
      }
    }
  ]
}

Return tool results in a later user message using tool_result blocks with the matching tool_use_id.

Thinking Blocks

Some Anthropic-compatible models can return thinking blocks. Preserve any returned signature when passing assistant thinking blocks into a later turn.

{
  "model": "anthropic/claude-opus-4.7",
  "max_tokens": 2048,
  "thinking": {
    "type": "enabled",
    "budget_tokens": 1024
  },
  "messages": [
    {
      "role": "user",
      "content": "Think carefully, then answer in one sentence."
    }
  ]
}

Response

{
  "id": "msg_xxx",
  "type": "message",
  "role": "assistant",
  "model": "anthropic/claude-opus-4.7",
  "content": [
    {
      "type": "text",
      "text": "Hello!"
    }
  ],
  "stop_reason": "end_turn",
  "usage": {
    "input_tokens": 10,
    "output_tokens": 5
  },
  "credit": 1.5
}

Streaming SSE

event: message_start
data: {"type":"message_start","message":{"id":"msg_xxx","type":"message","role":"assistant","content":[]}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"Hello"}}

event: message_delta
data: {"type":"message_delta","delta":{"stop_reason":"end_turn"},"usage":{"output_tokens":5}}

event: credit
data: {"credit":1.5}

event: message_stop
data: {"type":"message_stop"}

Code Examples

import anthropic

client = anthropic.Anthropic(
    base_url="https://shark.ai/api/v1",
    api_key="<api-key>",
)

response = client.messages.create(
    model="anthropic/claude-opus-4.7",
    max_tokens=2048,
    messages=[{"role": "user", "content": "Hello!"}],
)
print(response.content[0].text)

See model list for models that advertise Anthropic API support.