OpenAI-compatible API quickstart: Responses and Chat Completions

Three steps to your first GPT call

1. 01Create a free account and generate one API key (it looks like sk-pool-…) — the same key already covers Claude models too.
2. 02Point your client at https://openai.api.apitoken.sale/v1 and authenticate with Authorization: Bearer — not x-api-key; that header belongs to the Anthropic surface.
3. 03Confirm the enabled models with GET /v1/models, then send a Responses request.

curl https://openai.api.apitoken.sale/v1/responses \
  -H "Authorization: Bearer $APITOKEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-5.6-sol",
    "input": "Reply with exactly: connected"
  }'

See also: Run Codex CLI on apiToken.sale

Use the official OpenAI SDK

The official SDKs work unchanged — only base_url and the key change. Keep the key in a server-side environment variable in production.

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["APITOKEN_API_KEY"],
    base_url="https://openai.api.apitoken.sale/v1",
)

response = client.responses.create(
    model="gpt-5.6-sol",
    input="Reply with exactly: connected",
)
print(response.output_text)

Chat Completions is served on the same host if your client expects it — the model ID and key stay the same.

completion = client.chat.completions.create(
    model="gpt-5.6-sol",
    messages=[{"role": "user", "content": "Hello"}],
)
print(completion.choices[0].message.content)

Which GPT models are available

The served set is pinned and priced in the engine; GET /v1/models is always the live answer. Today the line covers three GPT-5.6 tiers and two previous-generation models:

• Reasoning effort is adjustable per request — none through xhigh on every model, plus max on the GPT-5.6 line.
• Every model accepts text and image input and streams over SSE on both Responses and Chat Completions.
• Requests above 272K input tokens bill at OpenAI long-context rates: 2× input and 1.5× output on the whole request.
• Your B2C discount applies here exactly as it does to Claude usage — one balance, one tier, 60–70% off official spend.

Full per-model specs and discounted prices →

What the endpoint does and does not cover

This is an independent OpenAI-compatible service, not the OpenAI Platform. It deliberately serves text generation only: model discovery, Responses, and Chat Completions with streaming and image input. Audio, files, realtime, assistants, batches and fine-tuning are not available.

Errors come in the OpenAI envelope — {"error":{"message","type","param","code"}}. A 401 means the key or the auth header is wrong (use Bearer, not x-api-key), a 402 means the shared prepaid balance needs a top-up, and a 404 means the model ID is not enabled — check GET /v1/models.

Frequently asked questions

Related guides