$5
50 API Tokens each month
Five-minute first-request tutorial
1
Subscribe to a plan
Open MagnetAPI.org on RapidAPI, choose a plan, and copy the application key from RapidAPI's request console.
2
Set two environment variables
Use your RapidAPI application key and the Marketplace host. Do not add a MagnetAPI System Token.
RAPIDAPI_KEY=your_rapidapi_application_key RAPIDAPI_HOST=magnetapi-org.p.rapidapi.com
3
Run the non-billable sandbox request
This confirms authentication and response shape without consuming an API Token or starting the 20-minute trial.
curl --request POST \
--url https://magnetapi-org.p.rapidapi.com/v1/responses \
--header "Content-Type: application/json" \
--header "X-RapidAPI-Key: $RAPIDAPI_KEY" \
--header "X-RapidAPI-Host: magnetapi-org.p.rapidapi.com" \
--data '{"model":"magnetapi-sandbox","input":"Verify my integration."}'4
Keep the request ID
Every response includes X-MagnetAPI-Request-ID. It is the fastest way for support to trace a request without storing your prompt or output.
5
Switch to a production model
Replace magnetapi-sandbox with a supported model ID. The 20-minute trial begins only after the first successful production generation.
Plans and quota calculator
One successful non-sandbox generation consumes one API Token after the trial. Failed requests and non-generation endpoints consume none.
$15
150 API Tokens each month
$50
500 API Tokens each month
$200
2,000 API Tokens each month
The estimate uses a 30-day month and excludes the one-time trial.
Estimated monthly need150 API Tokens
Developer is the closest matching plan.
Customer workflow case studies
These anonymized workflows summarize common integration patterns without exposing customer prompts or response content.
One base-URL change
A backend already using an OpenAI-compatible client changed its host and headers, validated with the sandbox model, then moved to production without redesigning its message pipeline.
Streaming with support diagnostics
An editor integration kept its streaming interface while adding the response request ID to logs, making intermittent support cases traceable without recording source code in the usage ledger.
Predictable monthly capacity
A small development team used the plan calculator and usage alerts to choose a monthly quota, then monitored endpoint latency and remaining API Tokens from one dashboard.
Test before spending quota
A prototype used the sandbox model in CI to verify JSON and streaming contracts. Production model calls were enabled only after those non-billable checks passed.
Framework starter repositories
Each download is a clean, self-contained starter with an environment template, setup README, non-billable sandbox default, and request-ID handling.
Node.js REST wrapper using native fetch.
Download ZIPAsync Python service with HTTPX.
Download ZIPMinimal Django JSON endpoint.
Download ZIPServer-side route plus a small test UI.
Download ZIPPostman workspace import
Fork the maintained public collection into your Postman workspace with one click, then set the rapidapiKey variable once. The collection saves the latest request ID automatically.
Short practical tutorials
Define a function, detect the tool call, execute it in your application, and return the result to the model.
Open tutorialRead OpenAI-compatible and Anthropic-compatible server-sent events without buffering the full answer.
Open tutorialLanguage examples and support tools
Create a private seven-day usage-dashboard link with POST /v1/rapidapi/dashboard-link. That dashboard provides quota alerts, endpoint performance, and copyable request IDs.