Responses

Creates a model response in the OpenAI Responses format. Codex CLI and newer OpenAI SDK code use it. Every free model answers here, streamed or not, with tool calls.

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

Request

curl

curl https://api.freetheai.org/v1/responses \
  -H "Authorization: Bearer $FTAI_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "fta/zai/glm-5.3",
    "instructions": "Answer in one sentence.",
    "input": "Why is the sky blue?"
  }'

Python

import os
from openai import OpenAI

client = OpenAI(base_url="https://api.freetheai.org/v1", api_key=os.environ["FTAI_KEY"])

response = client.responses.create(
    model="fta/zai/glm-5.3",
    instructions="Answer in one sentence.",
    input="Why is the sky blue?",
)
print(response.output_text)

TypeScript

import OpenAI from 'openai'

const client = new OpenAI({ baseURL: 'https://api.freetheai.org/v1', apiKey: process.env.FTAI_KEY })

const response = await client.responses.create({
  model: 'fta/zai/glm-5.3',
  instructions: 'Answer in one sentence.',
  input: 'Why is the sky blue?',
})
console.log(response.output_text)

Body parameters

Parameter	Type	Description
model	string, required	A model ID from GET /v1/models.
input	string or array	The input: a string, or a list of input items (messages with input_text and input_image content, or function_call_output items).
instructions	string	A system-style instruction for this response.
stream	boolean	Return server-sent events. Default false.
max_output_tokens	integer	The most tokens to generate. Send 0 to ask for the model's full output limit.
temperature, top_p	number	Sampling, where the model supports it.
tools	array	Function tools ({"type": "function", "name": ..., "parameters": ...}), or {"type": "web_search"} for the fta/dog/ and fta/lion/ models.
tool_choice	string or object	auto, none, required, or a specific function.
reasoning	object	For example {"effort": "low"}, where the model supports it.

Other Responses fields are passed to the model when it supports them.

Conversation state

Send the conversation in input each time. Server-side conversation storage (previous_response_id and store) is not something to rely on: FreeTheAI does not store prompts or replies.

Response

{
  "id": "resp_7c1e...",
  "object": "response",
  "model": "fta/zai/glm-5.3",
  "output": [
    {
      "type": "message",
      "role": "assistant",
      "content": [
        { "type": "output_text", "text": "Air molecules scatter short blue wavelengths of sunlight more than long red ones." }
      ]
    }
  ],
  "usage": { "input_tokens": 22, "output_tokens": 17, "total_tokens": 39 }
}

Field	Description
output	Output items: message items with output_text content, function_call items, and reasoning items for models that think.
output_text	SDK helper that joins the text of every output_text part.
usage	input_tokens, output_tokens, total_tokens.

Streaming

With "stream": true, the response is a stream of typed events: response.created, then response.output_text.delta events with the text, and response.completed with the full response and usage.

stream = client.responses.create(model="fta/zai/glm-5.3", input="Tell me a joke.", stream=True)
for event in stream:
    if event.type == "response.output_text.delta":
        print(event.delta, end="", flush=True)

Tool calls

import json

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

items = [{"role": "user", "content": "Weather in Oslo?"}]
response = client.responses.create(model="fta/zai/glm-5.3", input=items, tools=tools)

for item in response.output:
    if item.type == "function_call":
        args = json.loads(item.arguments)
        items.append(item.model_dump(exclude_none=True))
        items.append({
            "type": "function_call_output",
            "call_id": item.call_id,
            "output": json.dumps({"city": args["city"], "temperature_c": 6}),
        })

final = client.responses.create(model="fta/zai/glm-5.3", input=items, tools=tools)
print(final.output_text)

Codex CLI

Codex uses this endpoint. A tested config.toml is in the Codex CLI guide.

Errors

The same statuses as Chat Completions. Every message: Errors.