Anthropic Messages

Creates a message in the Anthropic Messages format. Claude Code and the Anthropic SDKs use it. Every free model answers here, streamed or not, with tool use.

POSThttps://api.freetheai.org/v1/messages

The Anthropic SDK and Claude Code take the base URL without /v1: https://api.freetheai.org.

Request

curl

curl https://api.freetheai.org/v1/messages \
  -H "x-api-key: $FTAI_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "fta/zai/glm-5.3",
    "max_tokens": 512,
    "system": "You are a concise assistant.",
    "messages": [{"role": "user", "content": "Give me a name for a grey cat."}]
  }'

Python

# pip install anthropic
import os
import anthropic

client = anthropic.Anthropic(
    base_url="https://api.freetheai.org",
    api_key=os.environ["FTAI_KEY"],
)

message = client.messages.create(
    model="fta/zai/glm-5.3",
    max_tokens=512,
    system="You are a concise assistant.",
    messages=[{"role": "user", "content": "Give me a name for a grey cat."}],
)
print(message.content[0].text)

TypeScript

// npm install @anthropic-ai/sdk
import Anthropic from '@anthropic-ai/sdk'

const client = new Anthropic({
  baseURL: 'https://api.freetheai.org',
  apiKey: process.env.FTAI_KEY,
})

const message = await client.messages.create({
  model: 'fta/zai/glm-5.3',
  max_tokens: 512,
  system: 'You are a concise assistant.',
  messages: [{ role: 'user', content: 'Give me a name for a grey cat.' }],
})
const first = message.content[0]
console.log(first.type === 'text' ? first.text : first)

Authorization: Bearer ftai_... works in place of x-api-key. Send the anthropic-version header your client normally sends; the SDKs do it for you.

Body parameters

Parameter	Type	Description
model	string, required	A model ID from GET /v1/models, such as fta/zai/glm-5.3. Anthropic model names do not work here.
max_tokens	integer, required	The most tokens to generate.
messages	array, required	Alternating user and assistant messages. content is a string or an array of content blocks (text, image, tool_use, tool_result).
system	string or array	The system prompt, as a string or text blocks.
stream	boolean	Return Anthropic server-sent events. Default false.
temperature, top_p, top_k	number	Sampling, where the model supports it.
stop_sequences	array	Text that ends the reply.
tools	array	Tools with name, description, and input_schema. {"type": "web_search_20250305", "name": "web_search"} turns on web search for fta/dog/ and fta/lion/ models.
tool_choice	object	{"type": "auto"}, {"type": "any"}, or {"type": "tool", "name": ...}.
thinking	object	{"type": "disabled"} turns reasoning off for models that think.

Other Anthropic fields are passed to the model when it supports them. Features that exist only in Claude models may not work on other models.

Response

{
  "id": "msg_01H...",
  "type": "message",
  "role": "assistant",
  "model": "fta/zai/glm-5.3",
  "content": [
    { "type": "text", "text": "How about Ash?" }
  ],
  "stop_reason": "end_turn",
  "usage": { "input_tokens": 24, "output_tokens": 6 }
}

Field	Description
content	Content blocks: text, tool_use, and thinking blocks for models that think.
stop_reason	end_turn, max_tokens, stop_sequence, or tool_use.
usage	input_tokens and output_tokens.

Streaming

With "stream": true you get Anthropic's events: message_start, content_block_start, content_block_delta, content_block_stop, message_delta, and message_stop. The SDK helpers work as usual:

with client.messages.stream(
    model="fta/zai/glm-5.3",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Write a limerick about tea."}],
) as stream:
    for text in stream.text_stream:
        print(text, end="", flush=True)

Tool use

See the Anthropic SDK page for a complete tool-use loop.

Claude Code

Claude Code sends this format to whatever ANTHROPIC_BASE_URL names. A tested settings.json is in the Claude Code guide.

Errors

The same statuses as Chat Completions, in FreeTheAI's error format. Every message: Errors.