Documentation

MagicAI API

MagicAI is an OpenAI-compatible gateway. The base URL is https://magicai.lol/v1; any client that lets you set a base URL and API key works without code changes.

Quickstart

Create an account and add credit under Dashboard → Credits (minimum $5).
Create a key under Dashboard → API Keys and copy it (it starts with sk-).
Send a request:

curl https://magicai.lol/v1/chat/completions \
  -H "Authorization: Bearer $MAGICAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "grok-4.7",
    "messages": [{"role": "user", "content": "Explain DNS in one sentence."}]
  }'

Authentication

Send your key as a bearer token on every request:

Authorization: Bearer sk-your-key

Keys draw from your account's shared credit balance. Treat them like passwords: keep them server-side, never ship them in client apps, and delete any key you suspect has leaked from the dashboard — deletion takes effect immediately.

Chat completions

POST /v1/chat/completions accepts the OpenAI Chat Completions request body.

Field	Type	Notes
model	string	Required. A live model id, e.g. grok-4.7 or gpt-6-luna.
messages	array	Required. Objects with role (system \| user \| assistant \| tool) and content.
stream	boolean	Return server-sent events instead of one JSON body.
max_tokens	integer	Upper bound on generated tokens.
temperature / top_p	number	Sampling controls.
stop	string \| array	Sequences that end generation.
tools / tool_choice	array / string	Function calling, on models that support it.
response_format	object	JSON or schema-constrained output, on models that support it.
reasoning_effort	string	For reasoning models that accept it (e.g. low, medium, high).

Fields are forwarded to the upstream provider; optional fields a model doesn't support may be ignored or rejected by that provider. The response follows the OpenAI format and includes a usage object with prompt_tokens and completion_tokens — the numbers you are billed for.

{
  "id": "…",
  "object": "chat.completion",
  "model": "grok-4.7",
  "choices": [{ "index": 0, "message": { "role": "assistant", "content": "…" }, "finish_reason": "stop" }],
  "usage": { "prompt_tokens": 18, "completion_tokens": 42, "total_tokens": 60 }
}

Streaming

Set "stream": true. You receive text/event-stream chunks, each a data: {…} line with a choices[0].delta, ending with data: [DONE]. Streaming is the best way to show output early; the time to first token shown on model pages is measured on streaming requests.

curl -N https://magicai.lol/v1/chat/completions \
  -H "Authorization: Bearer $MAGICAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"grok-4.7","stream":true,"messages":[{"role":"user","content":"Count to five"}]}'

Listing models

GET /v1/models (with your key) returns the model ids your key can call, in OpenAI list format. For public, key-less information, the model catalog shows live models with their per-token prices and measured speed.

curl https://magicai.lol/v1/models -H "Authorization: Bearer $MAGICAI_API_KEY"

Errors & rate limits

Errors use the OpenAI shape: {"error": {"message": "…", "type": "…", "code": "…"}}.

401	Missing, invalid or deleted API key.
403	Insufficient credit (code insufficient_user_quota). Top up and retry.
400	Malformed request or a parameter the upstream model rejects.
429	Too many requests — usually an upstream provider limit. Retry with exponential backoff.
5xx	Upstream failure after the gateway's own retries, or the model is not available. Retry later or try another model.

The gateway retries failed upstream attempts up to 2 times before returning an error. MagicAI does not currently set its own per-key request cap; limits you hit come from upstream providers. Sign-up and login are rate-limited per IP to block abuse.

Billing & credits

Credit is bought in US dollars through Stripe Checkout, minimum $5. There is no subscription.
Each request costs prompt_tokens × input price + completion_tokens × output price, using the per-1M-token rates on the model page.
Your balance, every request and its cost are visible under Dashboard → Usage.
When the balance reaches zero, requests return 403 until you top up. Nothing is charged automatically.

See Pricing and the Refund Policy.

SDKs & tools

Use the official OpenAI SDKs — just override the base URL.

# Python
pip install openai
client = OpenAI(base_url="https://magicai.lol/v1", api_key=os.environ["MAGICAI_API_KEY"])

// Node.js
npm install openai
const client = new OpenAI({ baseURL: "https://magicai.lol/v1", apiKey: process.env.MAGICAI_API_KEY });

# Environment variables respected by many OpenAI-compatible tools
export OPENAI_BASE_URL=https://magicai.lol/v1
export OPENAI_API_KEY=sk-your-magicai-key

Tools that accept a custom OpenAI-compatible endpoint can usually be pointed at MagicAI the same way.