Messages (Anthropic)

#Use it when

• Your code already targets the Anthropic Messages API
• You need Anthropic-style request and response objects
• You want Claude-native tooling and streaming behavior

#Code examples

1curl -X POST https://api.navy/v1/messages \
2  -H "x-api-key: sk-navy-YOUR_KEY" \
3  -H "Content-Type: application/json" \
4  -H "anthropic-version: 2023-06-01" \
5  -d '{
6    "model": "claude-sonnet-4-20250514",
7    "max_tokens": 1024,
8    "messages": [
9      {"role": "user", "content": "Draft a deployment checklist."}
10    ]
11  }'

#Parameters

• model (string, required) — Anthropic model ID such as claude-sonnet-4, claude-opus-4.6, claude-haiku-4.5
• messages (array, required) — Array with role ("user" or "assistant") and content
• max_tokens (integer, required) — Maximum tokens to generate (required by Anthropic format)
• system (string or array, optional) — System prompt as a string or array of text blocks
• stream (boolean, optional) — Enable streaming via Anthropic SSE format
• temperature (number, optional) — Sampling temperature (0.0–1.0)
• top_p (number, optional) — Nucleus sampling threshold
• top_k (integer, optional) — Top-K sampling parameter
• stop_sequences (array, optional) — Array of stop sequences
• tools (array, optional) — Anthropic-format tool definitions
• tool_choice (object, optional) — { type: "auto" }, { type: "any" }, or { type: "tool", name: "..." }

#Authentication

This endpoint accepts your NavyAI API key via either header:

• x-api-key: sk-navy-YOUR_KEY (Anthropic style)
• Authorization: Bearer sk-navy-YOUR_KEY (OpenAI style)

#Streaming

Set "stream": true to receive Anthropic-style SSE events such as message_start, content_block_start, content_block_delta, content_block_stop, message_delta, and message_stop.

1import anthropic
2
3client = anthropic.Anthropic(
4    api_key="sk-navy-YOUR_KEY",
5    base_url="https://api.navy"
6)
7
8with client.messages.stream(
9    model="claude-sonnet-4-20250514",
10    max_tokens=1024,
11    messages=[{"role": "user", "content": "Stream a haiku about the sea."}]
12) as stream:
13    for text in stream.text_stream:
14        print(text, end="", flush=True)

#Tool use

Provide tools and the model can emit tool_use content blocks. Reply with a tool_result block keyed by the tool's id on the next turn.

1tools = [{
2    "name": "get_weather",
3    "description": "Get the current weather for a city",
4    "input_schema": {
5        "type": "object",
6        "properties": {"city": {"type": "string"}},
7        "required": ["city"]
8    }
9}]
10
11message = client.messages.create(
12    model="claude-sonnet-4-20250514",
13    max_tokens=1024,
14    tools=tools,
15    messages=[{"role": "user", "content": "What's the weather in Tokyo?"}]
16)
17print(message.content)

#Vision

Send images using Anthropic's native image content block (URL or base64) inside a user message. Use Anthropic's documented shape — NavyAI does not transform it.

#Notes

• This endpoint is designed for Anthropic-native clients
• Streaming uses Anthropic SSE formatting
• Vision and tools are supported in the native shape
• If you call this endpoint with a non-Anthropic model (e.g. gpt-5), NavyAI adapts the request and translates the response back into Anthropic shape so the same SDK works across providers