Yes. The Free tier is 600 requests a month at 100 requests a minute, across every endpoint in the catalog, with no credit card. Nothing about it expires.

What happens when I hit the quota?

Requests return 429 with the limit, what you have used, and which window you tripped — nothing is queued, and nothing is charged. The monthly quota resets on the 1st (UTC). There is no rollover: unused requests do not carry into the next month.

Do failed requests use up my quota?

A 5xx does not. If a source breaks on our side, the request is excluded from your monthly count. A 4xx does count — a wrong parameter is still a request we answered. Per-minute throttling counts everything, so a retry loop is still throttled.

Which tier do I need?

Start on Free. Every response carries the remaining quota in its headers, and the dashboard reads the same number enforcement uses, so you will see the ceiling coming rather than discovering it.

How is Zapi different from RapidAPI or Apify?

Zapi focuses on a consistent REST contract — one API key, one response format, one error pattern for every source. No need to learn per-provider quirks like on traditional marketplaces, and no need to set up scraping workflows like on Apify.

What about the legality of accessing public data?

Zapi only accesses data that is publicly available without a login. Responsibility for how it is ultimately used stays with the developer — make sure your use case complies with the original source's ToS and the regulations in your operating region.

Is there an official SDK?

One: zpi-sdk, for TypeScript and JavaScript. It carries a typed error hierarchy, retries only what is safe to retry, submits bulk jobs, and verifies webhook signatures. Everything else is plain REST — cURL or any HTTP client works.

Can I request a source that isn't in the catalog yet?

Yes, through the community Discord, on any tier.

Why are Pro and Ultra marked Soon?

Checkout is not open yet. The limits and features listed on those tiers are already built and enforced — what is pending is billing, not the product.

ChatGPT (chatgpt.com)

ChatGPT (mode anonim, tanpa login) dari OpenAI. Endpoint kompatibel OpenAI Chat Completions (stream & non-stream).

llms.txt

Chat Completions

Kirim percakapan (OpenAI-format messages) ke ChatGPT anonim. Dukung stream (SSE delta) & non-stream.

POST/v1/ai:chatgpt/chatno cache3 paramschecking key…

Parameters

Query

streamboolean

Stream balasan via SSE (OpenAI delta, token real-time). Default false

Body · JSON

messagesarrayrequired

Valid JSON

OpenAI-format message history. Multi-turn penuh: system + riwayat user/assistant + pertanyaan terbaru semua dikirim sebagai konteks. System message dihormati untuk behavior (gaya/peran), bukan spoof identitas. VISION (gambar): kirim part image_url di content user, mis. content:[{"type":"text","text":"Baca teks di gambar"},{"type":"image_url","image_url":{"url":"https://.../foto.png"}}] (URL http(s) atau data:image base64 dua-duanya jalan). Vision REAL (OCR akurat). Model & tier anonim fixed (tak bisa dipilih).

modelstring

text/plain

Diabaikan (anonymous tier fixed). Diterima untuk kompat SDK OpenAI.

Response

EXAMPLE

{
  "id": "chatcmpl-QR1esXzhQuXQeEEuu5hCj",
  "model": "chatgpt",
  "usage": {
    "total_tokens": 116,
    "prompt_tokens": 60,
    "completion_tokens": 56
  },
  "object": "chat.completion",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "401 berarti klien belum terautentikasi atau kredensialnya salah sehingga akses ditolak karena harus login terlebih dahulu. 403 berarti klien sudah terautentikasi tetapi tidak memiliki izin untuk mengakses resource tersebut."
      },
      "finish_reason": "stop"
    }
  ],
  "created": 1782118631
}

Code

Chat Completions

index.ts

// npm i zpi-sdk
import { ZpiClient } from "zpi-sdk";

const client = new ZpiClient({ apiKey: "zpi_xxxxxxxxxxxxxxxxxxxxxxxx" });

const data = await client.run("ai:chatgpt", "chat", {
  "messages": [
    {
      "role": "system",
      "content": "Kamu adalah asisten yang ramah dan ringkas. Selalu jawab dalam bahasa Indonesia."
    },
    {
      "role": "user",
      "content": "Halo, aku lagi belajar HTTP."
    },
    {
      "role": "assistant",
      "content": "Tentu! HTTP itu protokol request-response. Mau mulai dari mana?"
    },
    {
      "role": "user",
      "content": "Jelaskan beda 401 dan 403 dalam satu kalimat masing-masing."
    }
  ],
  "stream": false
});
console.log(data);

Recommended · typed errors, safe retries and bulk jobsSDK reference

Full reference

1 endpoint · plain text

POST

Chat Completions/v1/ai:chatgpt/chat

3 params

Parameters

messagesarrayrequiredin body: OpenAI-format message history. Multi-turn penuh: system + riwayat user/assistant + pertanyaan terbaru semua dikirim sebagai konteks. System message dihormati untuk behavior (gaya/peran), bukan spoof identitas. VISION (gambar): kirim part image_url di content user, mis. content:[{"type":"text","text":"Baca teks di gambar"},{"type":"image_url","image_url":{"url":"https://.../foto.png"}}] (URL http(s) atau data:image base64 dua-duanya jalan). Vision REAL (OCR akurat). Model & tier anonim fixed (tak bisa dipilih).
streambooleanoptional: Stream balasan via SSE (OpenAI delta, token real-time). Default false
modelstringoptionalin body: Diabaikan (anonymous tier fixed). Diterima untuk kompat SDK OpenAI.

Request

curl -X POST "https://api.zapi.ink/v1/ai:chatgpt/chat?stream=false" \
  -H "x-api-key: zpi_xxxxxxxxxxxxxxxxxxxxxxxx"

Example response

{
  "id": "chatcmpl-QR1esXzhQuXQeEEuu5hCj",
  "model": "chatgpt",
  "usage": {
    "total_tokens": 116,
    "prompt_tokens": 60,
    "completion_tokens": 56
  },
  "object": "chat.completion",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "401 berarti klien belum terautentikasi atau kredensialnya salah sehingga akses ditolak karena harus login terlebih dahulu. 403 berarti klien sudah terautentikasi tetapi tidak memiliki izin untuk mengakses resource tersebut."
      },
      "finish_reason": "stop"
    }
  ],
  "created": 1782118631
}

{ "id": "chatcmpl-QR1esXzhQuXQeEEuu5hCj", "model": "chatgpt", "usage": { "total_tokens": 116, "prompt_tokens": 60, "completion_tokens": 56 }, "object": "chat.completion", "choices": [ { "index": 0, "message": { "role": "assistant", "content": "401 berarti klien belum terautentikasi atau kredensialnya salah sehingga akses ditolak karena harus login terlebih dahulu. 403 berarti klien sudah terautentikasi tetapi tidak memiliki izin untuk mengakses resource tersebut." }, "finish_reason": "stop" } ], "created": 1782118631 }

// npm i zpi-sdk import { ZpiClient } from "zpi-sdk"; const client = new ZpiClient({ apiKey: "zpi_xxxxxxxxxxxxxxxxxxxxxxxx" }); const data = await client.run("ai:chatgpt", "chat", { "messages": [ { "role": "system", "content": "Kamu adalah asisten yang ramah dan ringkas. Selalu jawab dalam bahasa Indonesia." }, { "role": "user", "content": "Halo, aku lagi belajar HTTP." }, { "role": "assistant", "content": "Tentu! HTTP itu protokol request-response. Mau mulai dari mana?" }, { "role": "user", "content": "Jelaskan beda 401 dan 403 dalam satu kalimat masing-masing." } ], "stream": false }); console.log(data);

ChatGPT (chatgpt.com)

Chat Completions

Code

Full reference

Parameters

Request

Example response

Copilot AI

Felo AI

Chatex

Dolphin AI

Google AI (g.ai)

Bypass AI (writecream.com)

ChatGPT (chatgpt.com)

Chat Completions

Code

Full reference

Parameters

Request

Example response

Copilot AI

Felo AI

Chatex

Dolphin AI

Google AI (g.ai)

Bypass AI (writecream.com)

Chat Completions

Code

Full reference

Parameters

Request

Example response

Related APIs

Copilot AI

Felo AI

Chatex

Dolphin AI

Google AI (g.ai)

Bypass AI (writecream.com)

Chat Completions

Code

Full reference

Parameters

Request

Example response

Related APIs

Copilot AI

Felo AI

Chatex

Dolphin AI

Google AI (g.ai)

Bypass AI (writecream.com)