> Parel documentation. Index of every page: https://parel.si/docs/llms.txt. This page in HTML: https://parel.si/docs/tools/search_patents_backup

# search_patents_backup

> A second source for search_patents: google Patents results for a query: title, number, assignee, inventors, dates and abstract. ₹0.20 per call; a result that finds nothing is ₹0.

## What it does

A second source for [`search_patents`](https://parel.si/docs/tools/search_patents.md), from a different data provider: google Patents results for a query: title, number, assignee, inventors, dates and abstract. The answer is this provider's own, so its field names differ from [`search_patents`](https://parel.si/docs/tools/search_patents.md)'s: read the example answer, not [`search_patents`](https://parel.si/docs/tools/search_patents.md)'s docs.

Use it when [`search_patents`](https://parel.si/docs/tools/search_patents.md) fails; [`search_patents`](https://parel.si/docs/tools/search_patents.md) stays the first choice. It reads the same platform through another provider, so it helps when the first one is down or blocked, not when the platform has nothing. It is a separate call at its own price: it is billed per call when it answers; a failed or empty call is free.

India: Indian patents appear with their IN numbers alongside the rest.

## Inputs

Generated from the contract captured from staging on 2026-11-18.

| Field | Type | Required | Description |
|---|---|---|---|
| `q` | string | yes | e.g. "transformer neural network" |
| `page` | integer | no | default 1; min 1 |
| `autocorrect` | boolean | no | default true |

## What drives the cost

Flat price per call. A result that finds nothing is `class: "ok_free"`, ₹0. The timeout is 30 seconds.

## Price today

| | |
|---|---|
| Price today | **₹0.20** per call (`cost_paise: 20`) at ₹93.93 per US dollar on 17 November 2026 |
| Hold before the call | ₹0.38 (`hold_paise: 38`); when the call settles the real price is charged and the unused part is released |
| ₹0 when | it fails, or the provider answers but there is nothing in it (an empty list, no email, no page): `class: "ok_free"`, ₹0 |
| Timeout | 30 s, then `upstream_timeout` (₹0, retryable) |
| Over MCP | `run_tool` with `tool: "search_patents_backup"` |
| Over REST | `POST https://api.parel.si/v1/run` with `"tool": "search_patents_backup"` |
| Confirmation | not needed (the estimate is under the ₹200 threshold) |

## Request

```bash
curl https://api.parel.si/v1/run \
  -H "Authorization: Bearer $PAREL_KEY" \
  -H "Content-Type: application/json" \
  -d '{"tool":"search_patents_backup","input":{"q":"transformer neural network"}}'
```

## What comes back

`data`, as the live check of 2026-11-18 recorded it, trimmed:

```json
{
  "searchParameters": {
    "q": "transformer neural network",
    "type": "patents",
    "num": 10,
    "page": 1,
    "engine": "google"
  },
  "organic": [
    {
      "title": "Harmonizing composite images utilizing a transformer neural network",
      "snippet": ", wherein utilizing the second neural network branch comprises utilizing one or more transformer neural network layers to extract the global information of the background. 3. The non-transitory computer-readable medium of claim 2 , wherein…",
      "link": "https://patents.google.com/patent/US12165284B2/en",
      "priorityDate": "2022-03-21",
      "filingDate": "2022-03-21",
      "grantDate": "2024-12-10",
      "publicationDate": "2024-12-10",
      "inventor": "He Zhang",
      "assignee": "Adobe Inc.",
      "publicationNumber": "US12165284B2",
      "language": "en",
      "thumbnailUrl": "https://patentimages.storage.googleapis.com/7a/22/b4/fba214148613df/US12165284-20241210-D00000.png",
      "pdfUrl": "https://patentimages.storage.googleapis.com/86/1b/3a/f258738f2792e2/US12165284.pdf",
      "figures": [
        {
          "imageUrl": "https://patentimages.storage.googleapis.com/57/c1/43/7f0746ee8823a7/US12165284-20241210-D00000.png",
          "thumbnailUrl": "https://patentimages.storage.googleapis.com/e9/73/77/428d402fa7f56a/US12165284-20241210-D00000.png"
        },
        {
          "imageUrl": "https://patentimages.storage.googleapis.com/3f/f4/5e/8aaf709014bd4d/US12165284-20241210-D00001.png",
          "thumbnailUrl": "https://patentimages.storage.googleapis.com/6a/ac/01/fe66e4be01a560/US12165284-20241210-D00001.png"
        },
        {
          "imageUrl": "https://patentimages.storage.googleapis.com/fc/d7/4d/afa706033eef33/US12165284-20241210-D00002.png",
          "thumbnailUrl": "https://patentimages.storage.googleapis.com/5b/5b/30/491343bd70bd40/US12165284-20241210-D00002.png"
        }
      ],
      "position": 1
    },
    {
      "title": "System, method, and computer program for transformer neural networks",
      "snippet": "14. A method comprising: implementing, by one or more processing device, a sequence of transformer neural networks, wherein implementing the sequence of trained transformer neural networks comprises implementing each transformer neural netw…",
      "link": "https://patents.google.com/patent/US12056592B2/en",
      "priorityDate": "2020-08-14",
      "filingDate": "2020-08-14",
      "grantDate": "2024-08-06",
      "publicationDate": "2024-08-06",
      "inventor": "Sanjeet Hajarnis",
      "assignee": "Eightfold AI Inc.",
      "publicationNumber": "US12056592B2",
      "language": "en",
      "thumbnailUrl": "https://patentimages.storage.googleapis.com/fb/15/34/3be33d61c66e9b/US12056592-20240806-D00000.png",
      "pdfUrl": "https://patentimages.storage.googleapis.com/cf/c7/18/df02c8e089124e/US12056592.pdf",
      "figures": [
        {
          "imageUrl": "https://patentimages.storage.googleapis.com/7d/9d/55/56dd3212160ad7/US12056592-20240806-D00000.png",
          "thumbnailUrl": "https://patentimages.storage.googleapis.com/07/f8/8d/115a3b669e955d/US12056592-20240806-D00000.png"
        },
        {
          "imageUrl": "https://patentimages.storage.googleapis.com/92/e2/ea/1a3accb7cf0b54/US12056592-20240806-D00001.png",
          "thumbnailUrl": "https://patentimages.storage.googleapis.com/94/6f/c8/962afc6d9116e8/US12056592-20240806-D00001.png"
        },
        {
          "imageUrl": "https://patentimages.storage.googleapis.com/8d/ca/33/cab4a5660d3fe8/US12056592-20240806-D00002.png",
          "thumbnailUrl": "https://patentimages.storage.googleapis.com/8c/92/9a/0a802cb0cc9606/US12056592-20240806-D00002.png"
        }
      ],
      "position": 2
    },
    {
      "title": "A point cloud data processing method based on transformer neural network",
      "snippet": "1. A point cloud data processing method based on a transformer neural network, characterized in that it includes: Step S1: Construct a three-dimensional object symmetry detection model. The three-dimensional object symmetry detection model…",
      "link": "https://patents.google.com/patent/CN113870160B/en",
      "priorityDate": "2021-09-10",
      "filingDate": "2021-09-10",
      "grantDate": "2024-02-27",
      "publicationDate": "2024-02-27",
      "inventor": "王旭",
      "assignee": "北京交通大学",
      "publicationNumber": "CN113870160B",
      "language": "en",
      "thumbnailUrl": "https://patentimages.storage.googleapis.com/",
      "pdfUrl": "https://patentimages.storage.googleapis.com/0a/f0/31/3b9b15751939ef/CN113870160B.pdf",
      "position": 3
    }
  ]
}
```

## Notes

- Second source for search_patents: call search_patents first, this one after a failure.
- Field names differ from search_patents's; inputs keep search_patents's names where they mean the same thing.

## Siblings

- [`search_patents`](https://parel.si/docs/tools/search_patents.md)
