# arXiv Papers Search API

> Search arXiv preprints by keyword, author, title or category: abstract, authors, dates, PDF link

- Vendor: `arxivpaperssearchapi` · version 1.0 · category Science · tags arxiv api, research papers api, preprints api, academic search api, physics papers api, ai papers api
- Datasource: custom HTTP routes
- Marketplace page: https://apiclay.com/api/arxivpaperssearchapi
- This document: https://apiclay.com/docs/arxivpaperssearchapi.md
- OpenAPI 3.1: https://apiclay.com/api/arxivpaperssearchapi/openapi.json
- MCP: connect https://api.apiclay.com/v1/mcp (see https://apiclay.com/llms.txt), then `apiclay_describe_api` with vendor `arxivpaperssearchapi`
- Updated: 2026-11-18

## About

## arXiv Papers Search API

Search arXiv, the open preprint server for physics, mathematics, computer science, statistics, biology and economics: each paper's title, abstract, authors, categories, submission and update dates, DOI and PDF link.

Data comes from **arxiv.org** (https://arxiv.org). Every route calls the source's own public API and returns it unchanged.

Typical flow: `/search?search_query=all:llm&amp;max_results=10&amp;sortBy=submittedDate&amp;sortOrder=descending`.

### Routes

#### GET `/search` - Search papers
| Param | In | Type | Required | Description |
|---|---|---|---|---|
| search_query | query | string | no | e.g. `all:llm`, `au:hinton`, `ti:transformer`, `cat:cs.CL`; combine with `AND`. |
| id_list | query | string | no | Comma-separated arXiv ids, e.g. `2401.00001`. |
| start | query | number | no | Offset. |
| max_results | query | number | no | Up to 2000. |
| sortBy | query | string | no | `relevance`, `lastUpdatedDate`, `submittedDate`. |
| sortOrder | query | string | no | `ascending` or `descending`. |

Returns: an Atom XML feed: `entry` elements with `id`, `title`, `summary`, `author`, `published`, `updated`, `category`, PDF `link`.

### Caveats

- The response is Atom XML, not JSON.
- arXiv asks for no more than one request every 3 seconds.
- Upstream responses can change over time. Verify fields before relying on them.
- Not affiliated with or endorsed by arxiv.org.

## Pricing

- Pay as you go: 0.0012 credits/request
- Trial: 1 free requests per buyer
- Billing order: trial requests -> subscribed plan requests -> pay as you go. 1 credit's USD value is shown on the marketplace page.

## HTTP routes

Only the public, buyer-facing parts of each route are listed; vendor-side credentials stay on the gateway.

### GET /search — Search papers
- Gateway: `GET https://api.apiclay.com/v1/api/arxivpaperssearchapi/search`
- Public query params: `search_query` (string), `id_list` (string), `start` (number), `max_results` (number), `sortBy` (string), `sortOrder` (string)
- Upstream proxy: ApiClay proxy paid by the vendor (data cap 2 MB per call). Fetched through an ApiClay proxy paid by the vendor; a response larger than the route's data cap fails with 502 and is refunded in full.

## Call example

```bash
curl -X GET "https://api.apiclay.com/v1/api/arxivpaperssearchapi/search" \
  -H "token: <YOUR_APICLAY_TOKEN>"
```

Index of every published API: https://apiclay.com/llms.txt
