Inference Cheap API

For agents — copy the integration prompt.

Connect

Client	Base URL
OpenAI SDKs	`https://openai.inference.cheap/v1`
Codex CLI / app	`https://codex.inference.cheap` — setup

Get an API key from the Inference Cheap team. Pass it as Authorization: Bearer YOUR_API_KEY or the SDK's api_key option. Keep keys server-side.

export INFERENCE_CHEAP_API_KEY="YOUR_API_KEY"
curl --fail-with-body https://openai.inference.cheap/v1/models \
  -H "Authorization: Bearer $INFERENCE_CHEAP_API_KEY"

First request

Install the openai package. Use a model returned by /v1/models.

Python

import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["INFERENCE_CHEAP_API_KEY"],
    base_url="https://openai.inference.cheap/v1",
)
response = client.responses.create(model="gpt-5.6-sol", input="Say hello.")
print(response.output_text)

JavaScript

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.INFERENCE_CHEAP_API_KEY,
  baseURL: "https://openai.inference.cheap/v1",
});
const response = await client.responses.create({
  model: "gpt-5.6-sol",
  input: "Say hello.",
});
console.log(response.output_text);

API endpoints

Paths are relative to https://openai.inference.cheap. All require an API key.

Operation	Endpoint
Responses / Chat Completions	`POST /v1/responses`, `POST /v1/chat/completions`
List / retrieve models	`GET /v1/models`, `GET /v1/models/{id}`
Retrieve / delete a stored response	`GET /v1/responses/{id}`, `DELETE /v1/responses/{id}`
Response inputs / background cancellation	`GET /v1/responses/{id}/input_items`, `POST /v1/responses/{id}/cancel`
Compact input / count tokens	`POST /v1/responses/compact`, `POST /v1/responses/input_tokens`
Conversations / files	`/v1/conversations`, `/v1/files`
Generate / edit images	Images API
Billing scope and multipliers	`GET /v1/sub2api/billing`

Request options

Models: /v1/models lists models available to your key, not prices. Model, tool, and reasoning support depend on your access.
Reasoning: omit for defaults. Responses uses reasoning: { effort: "high" }; Chat Completions uses reasoning_effort: "high".
Streaming: set stream: true. Read response.output_text.delta events; handle response.failed and response.incomplete. HTTP 200 alone is not success.
History: set store: true for later retrieval. Use previous_response_id or a conversation ID, not both; resend instructions for each response.
Pricing: ask the team for rates and usage. /v1/sub2api/billing returns scope and multipliers, not per-model prices.

Errors

Status	Action
`400` / `404`	Check the error, request fields, endpoint, model, and resource IDs.
`401` / `403`	Check your key and access.
`413`	Reduce the payload.
`429`	Reduce concurrency; respect `Retry-After`.
`5xx`	Retry with bounded backoff.

After a timeout, inspect a stored response before repeating the request.

For agents

Copy this Markdown prompt to your agent.

Integrate the API

# Integrate Inference Cheap

- Use the OpenAI SDK with base URL https://openai.inference.cheap/v1.
- Read INFERENCE_CHEAP_API_KEY from the project's secret store; never expose it
  in browser code or logs. Ask the Inference Cheap team if a key is missing.
- Call GET /v1/models before choosing a model; keep the user's choice if available.
- Start with Responses and default options. Preserve existing client settings.
- For streaming, check the final outcome; HTTP 200 is not proof of success.
- When testing is requested, make one small request and report its final result.
- Codex setup: https://docs.inference.cheap/codex
- Image generation: https://docs.inference.cheap/images

On this page