ChatGPT Scraper API

Collect ChatGPT answers, follow-up answers and citations for independent questions. Receive a task ID immediately after acceptance, then poll for progress and results.

Submit a task

POST https://api.goanyapi.com/api/v1/chatgpt-scraper/tasks

Authenticate with Authorization: Bearer YOUR_GOANYAPI_API_KEY. Send JSON with inputs (1–20 questions) and idempotency_key (8–128 letters, numbers, underscores or hyphens). Each question starts its own conversation; additional_prompt follows the first answer within that conversation. The target URL is fixed and is not an input parameter.

Countries and regions

country accepts only the supported uppercase two-letter codes below. Omit it or use an empty string for automatic location. Location does not restrict the prompt language. CN, HK and MO are unsupported. An unsupported code returns HTTP 400 before credits are reserved.

{
  "idempotency_key": "research_batch_001",
  "inputs": [
    {
      "prompt": "Compare project management tools",
      "country": "US",
      "index": 1,
      "web_search": true,
      "require_sources": true,
      "additional_prompt": "Which options offer a free plan?",
      "geolocation": "37.7749,-122.4194"
    },
    { "prompt": "What are the latest renewable energy trends?", "index": 2 }
  ]
}

A new valid submission returns HTTP 202 immediately after credits are reserved and the task is persisted, with data.task_id and status: "queued". It does not wait for processing or results. The ID confirms GoAnyAPI has accepted the task; it does not mean the answer is ready. Save it and use the query endpoint below. Retry a lost response with the same idempotency key and identical inputs to retrieve the same task without duplicate charges. Different inputs with the same key return HTTP 409. A new key creates a separately billed task. Reusing an already completed task can return HTTP 200 with its saved result. Keys are retained with tasks; do not reuse them after retention expires.

Query status and results

GET https://api.goanyapi.com/api/v1/chatgpt-scraper/tasks/{task_id}

Replace {task_id} with the ID returned by submission. Use the same authenticated account; another account cannot query your task. Poll at intervals of at least 5 seconds. queued and running are pending; succeeded and failed are terminal. A succeeded batch can include failed questions: inspect each results[].status. Every item contains its tracking index and either result or a public error.

Query response fields

Task states

result answer fields

All fields below may be missing, null or empty; empty fields alone do not mean failure.

Arrays may contain strings, numbers, booleans, null or objects. Objects retain only position, text, title, url, description, domain, icon, cited, name, price, currency, rating, address, latitude and longitude; these fields may also be absent or empty.

Query response JSON examples

All examples below use HTTP 200 and illustrate the structure rather than real tasks. Billing figures use the current configuration; actual task rates are fixed at creation.

{  "code": "ok",  "message": "ok",  "data": {    "task_id": "cgs_aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",    "status": "queued",    "created_at": "2026-11-10T00:00:00.000Z",    "expires_at": "2026-12-10T00:00:00.000Z",    "input_count": 1,    "credits_reserved": 3,    "credits_refunded": 0,    "credits_used": 0,    "poll_after_seconds": 5  }}

{  "code": "ok",  "message": "ok",  "data": {    "task_id": "cgs_aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",    "status": "running",    "created_at": "2026-11-10T00:00:00.000Z",    "expires_at": "2026-12-10T00:00:00.000Z",    "input_count": 1,    "credits_reserved": 3,    "credits_refunded": 0,    "credits_used": 0,    "poll_after_seconds": 5  }}

{  "code": "ok",  "message": "ok",  "data": {    "task_id": "cgs_aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",    "status": "succeeded",    "created_at": "2026-11-10T00:00:00.000Z",    "expires_at": "2026-12-10T00:00:00.000Z",    "input_count": 1,    "credits_reserved": 3,    "credits_refunded": 0,    "credits_used": 3,    "results": [      {        "index": 1,        "status": "succeeded",        "result": {          "index": 1,          "prompt": "Example question",          "answer_text": "Example answer.",          "citations": [],          "answer_html": "<p>Example answer.</p>"        }      }    ]  }}

{  "code": "ok",  "message": "ok",  "data": {    "task_id": "cgs_aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",    "status": "failed",    "created_at": "2026-11-10T00:00:00.000Z",    "expires_at": "2026-12-10T00:00:00.000Z",    "input_count": 1,    "credits_reserved": 3,    "credits_refunded": 3,    "credits_used": 0,    "results": [      {        "index": 1,        "status": "failed",        "error": {          "code": "collection_failed",          "message": "Unable to deliver a result for this question"        }      }    ]  }}

{  "code": "ok",  "message": "ok",  "data": {    "task_id": "cgs_aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa",    "status": "succeeded",    "created_at": "2026-11-10T00:00:00.000Z",    "expires_at": "2026-12-10T00:00:00.000Z",    "input_count": 2,    "credits_reserved": 6,    "credits_refunded": 3,    "credits_used": 3,    "results": [      {        "index": 1,        "status": "succeeded",        "result": {          "index": 1,          "prompt": "Example question",          "answer_text": "Example answer.",          "citations": [],          "answer_html": "<p>Example answer.</p>"        }      },      {        "index": 2,        "status": "failed",        "error": {          "code": "collection_failed",          "message": "Unable to deliver a result for this question"        }      }    ]  }}

Billing and retention

3 credits per question, deducted at submission; failed inputs are automatically refunded. Queries are free.

Credits are deducted for every input at submission. Insufficient balance prevents task creation. A background processing failure refunds all credits; partial failures refund the affected questions. Read the terminal task and each result to see failures and refunds. Successful questions are billed whether or not you view the answers. Sources, fields or citations being empty does not by itself make a result a failure. Queries and repeated result downloads are free. A follow-up is included in the per-input rate.

Processing continues without an open page or polling. Temporary network errors do not immediately fail or refund a task. Tasks not delivered within 24 hours are closed and refunded when next processed or queried; refunds may be delayed without subsequent activity. A later recovery does not charge you again. Completed results are retained for 30 days. Keep batches small: a task supports downloaded JSON up to 20 MB and a retained record up to 1.8 MB including task metadata; undeliverable oversized results are refunded. Expired results return HTTP 410 while the record exists, then HTTP 404 after cleanup. Refunds keep the original credits' expiration date.

Errors

HTTP 400: invalid JSON, unknown parameters, duplicate tracking indexes or invalid values. HTTP 401: invalid authentication. HTTP 402: insufficient credits. HTTP 404: task not found or inaccessible. HTTP 409: idempotency key conflict. HTTP 410: expired task. HTTP 429: rate limit. HTTP 503: service or billing configuration unavailable before acceptance. Background processing failures are reported by the query endpoint as status: "failed" and results[].error, with refunds in credits_refunded. The public envelope uses code, message and data.