Anthropic SDK

FreeTheAI answers the Anthropic Messages format for every free model, so the official Anthropic SDKs work with a different base URL, your FreeTheAI key, and an fta/ model ID.

• Base URL: https://api.freetheai.org (without /v1; the SDK adds /v1/messages)
• Key: your ftai_ key as the API key
• Model: an fta/ ID such as fta/zai/glm-5.3. Claude model names do not work.

Not Claude

These requests run on the model you name, not on Claude. Features that only exist in Claude models may not work.

Install and configure

Python

# pip install anthropic
import os
import anthropic

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

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,
})

The SDKs also read ANTHROPIC_BASE_URL and ANTHROPIC_API_KEY from the environment.

Create a message

Python

message = client.messages.create(
    model="fta/zai/glm-5.3",
    max_tokens=1024,
    system="You are a helpful assistant.",
    messages=[{"role": "user", "content": "Explain DNS in three sentences."}],
)
for block in message.content:
    if block.type == "text":
        print(block.text)

TypeScript

const message = await client.messages.create({
  model: 'fta/zai/glm-5.3',
  max_tokens: 1024,
  system: 'You are a helpful assistant.',
  messages: [{ role: 'user', content: 'Explain DNS in three sentences.' }],
})
for (const block of message.content) {
  if (block.type === 'text') console.log(block.text)
}

Streaming

Python

with client.messages.stream(
    model="fta/kimi/k3",
    max_tokens=2048,
    messages=[{"role": "user", "content": "Write a short poem about rain."}],
) as stream:
    for text in stream.text_stream:
        print(text, end="", flush=True)

TypeScript

const stream = client.messages.stream({
  model: 'fta/kimi/k3',
  max_tokens: 2048,
  messages: [{ role: 'user', content: 'Write a short poem about rain.' }],
})
stream.on('text', (text) => process.stdout.write(text))
await stream.finalMessage()

Tool use

import json

tools = [{
    "name": "get_weather",
    "description": "Get the current weather for a city.",
    "input_schema": {
        "type": "object",
        "properties": {"city": {"type": "string"}},
        "required": ["city"],
    },
}]

messages = [{"role": "user", "content": "Is it cold in Reykjavik right now?"}]

while True:
    message = client.messages.create(
        model="fta/zai/glm-5.3", max_tokens=1024, tools=tools, messages=messages,
    )
    messages.append({"role": "assistant", "content": message.content})
    if message.stop_reason != "tool_use":
        print("".join(b.text for b in message.content if b.type == "text"))
        break
    results = []
    for block in message.content:
        if block.type == "tool_use":
            weather = {"city": block.input["city"], "temperature_c": -2}  # your lookup here
            results.append({
                "type": "tool_result",
                "tool_use_id": block.id,
                "content": json.dumps(weather),
            })
    messages.append({"role": "user", "content": results})

Errors

The SDK raises anthropic.APIStatusError subclasses (AuthenticationError, PermissionDeniedError, NotFoundError, RateLimitError, ...). FreeTheAI's error body is the same on every endpoint; see Errors.

Claude Code

Claude Code uses the same endpoint. See the Claude Code guide for a tested settings.json.