Responses API

Overview

Token101 supports the OpenAI Responses API (/api/v1/responses), the same protocol used by OpenAI Codex CLI and other agent-oriented tools. The Responses API is designed for multi-step workflows where the model can reason, call tools, and produce structured outputs in a single request-response cycle.

Endpoint

POST https://token.ppthub.shop/api/v1/responses

Authentication

Include your Token101 API key in the Authorization header:

Authorization: Bearer sk-your-token101-api-key

Quick Example

curl

curl https://token.ppthub.shop/api/v1/responses \
  --request POST \
  --header "Content-Type: application/json" \
  --header "Authorization: Bearer $TOKEN101_API_KEY" \
  --data '{
    "model": "gpt-5.2",
    "input": "What are the three laws of robotics?"
  }'

Python (OpenAI SDK)

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["TOKEN101_API_KEY"],
    base_url="https://token.ppthub.shop/api/v1",
)

response = client.responses.create(
    model="gpt-5.2",
    input="What are the three laws of robotics?",
)

print(response.output_text)

Node.js / TypeScript (OpenAI SDK)

import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: process.env.TOKEN101_API_KEY!,
  baseURL: 'https://token.ppthub.shop/api/v1',
});

const response = await client.responses.create({
  model: 'gpt-5.2',
  input: 'What are the three laws of robotics?',
});

console.log(response.output_text);

Request Fields

The Responses API accepts the standard OpenAI Responses API request format. Key fields:

For the complete field reference, see the API Reference.

Streaming

Set stream: true to receive Server-Sent Events (SSE) as the response is generated:

response = client.responses.create(
    model="gpt-5.2",
    input="Explain quantum computing in simple terms.",
    stream=True,
)

for event in response:
    if event.type == "response.output_text.delta":
        print(event.delta, end="", flush=True)

Streaming events follow the OpenAI Responses API format — response.created, response.output_item.added, response.output_text.delta, response.completed, and others.

Supported Models

The following models support the Responses API endpoint:

See Supported Models for the full list including non-OpenAI models.

How It Works

When you send a request to /api/v1/responses:

1. Token101 validates your API key and checks rate limits and billing
2. The request body is forwarded to the upstream model provider by Token101's routing layer
3. The upstream provider processes the Responses API request natively
4. The response (streaming or non-streaming) passes back through Token101 to your client

Token101 preserves the full Responses API contract — tool calls, structured outputs, reasoning tokens, and streaming all work as expected with supported models.

Comparison with Other Endpoints

Known Limitations

• Server-side tool execution: Token101's Responses API is a passthrough adapter. Server-side tools (e.g., web search, file search) rely on the upstream provider's native support.
• Model availability: Only OpenAI models currently support the Responses API natively. Other models (Claude, Gemini, Qwen) should use the Messages or Chat Completions endpoints instead.
• Reasoning token billing: Reasoning tokens generated through the Responses API receive the same automatic discounts as other endpoints (Codex series 50% off, non-Codex GPT 25% off the output rate). See Billing for details.

Troubleshooting

401 Unauthorized

Check that your API key is set and valid. See Authentication for details.

404 Not Found on /api/v1/responses

Confirm your base URL is correct:

# Correct
export OPENAI_BASE_URL="https://token.ppthub.shop/api/v1"

# Wrong — do not include /responses in the base URL
export OPENAI_BASE_URL="https://token.ppthub.shop/api/v1/responses"

400 Invalid model

The model you specified may not support the Responses API. Use one of the model IDs listed in the Supported Models table above.

Model returns unexpected format

Some older OpenAI models may not fully support the Responses API format. Try using a newer model like gpt-5.2 or gpt-5.3-codex.

Next Steps

• Codex CLI Integration — use Codex CLI with Token101
• Supported Models — full list of available models
• Streaming — streaming guide for all endpoints
• API Reference — full endpoint documentation