Skip to main content

Messages

POSThttps://api.stdcmpt.com/v1/messages

The Anthropic Messages API. Claude Code and the Anthropic SDKs use this endpoint. For Claude Code, see the Claude Code guide.

Request​

Set the Anthropic SDK's base URL to https://api.stdcmpt.com (without /v1):

import os
from anthropic import Anthropic

client = Anthropic(
    base_url="https://api.stdcmpt.com",
    api_key=os.environ["STANDARDCOMPUTE_API_KEY"],
)

message = client.messages.create(
    model="anthropic/claude-standardcompute",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Review this function for bugs: ..."}],
)

print(message.content[0].text)

Authenticate with x-api-key or Authorization: Bearer.

Models​

  • anthropic/claude-standardcompute or standardcompute: smart routing.
  • An exact ID such as anthropic/claude-sonnet-5.5: that model only.
  • Older Claude Code names (sonnet, opus, haiku and legacy dated IDs): smart routing, so existing setups keep working.

With smart routing, the response's model field echoes the model you requested.

Supported features​

  • Streaming with standard Anthropic server-sent events.
  • Tool use with client-defined tools that have an input_schema. Anthropic server tools are ignored.
  • Images in message content, on vision models.
Not supported

The thinking parameter is ignored and responses don't include thinking blocks.

Count tokens​

POSThttps://api.stdcmpt.com/v1/messages/count_tokens

Returns an estimated input token count for a Messages request body:

{ "input_tokens": 1834 }

The count is an estimate based on request size, not an exact tokenizer count.