arXiv Search API for papers and preprints
Search arXiv over HTTP and receive the authors, the submission date, and the abstract in a single field, without parsing the Atom feed the native interface returns.
Endpoint
Results arrive in the same JSON shape as every other engine, so existing parsing continues to work.
The link points at the paper page on arxiv.org.
The snippet begins with the author list and the date, in ISO form.
The rest of the snippet is the paper abstract.
Useful for
Literature review / Research agents / Trend tracking
Preprints as JSON rather than Atom
Set `search_service` to `arxiv` and the results come from arXiv. The endpoint, the authentication, and the response shape are the same as every other engine, so no separate feed parser is needed.
Citation fields in the snippet
The `snippet` field follows the pattern `authors | date | abstract`, so a citation can be assembled without a second call.
ISO submission dates
The date arrives as `YYYY-MM-DD`, which sorts and filters without normalisation.
Abstracts, not just titles
The abstract is in the snippet, so a model can judge relevance before deciding whether to fetch the paper.
Standard Search parameters
`max_results` up to 50, plus `include_sites`, `exclude_sites`, and `time_range`.
What is an arXiv search API?
An arXiv search API is an HTTP interface that returns arXiv papers as structured results. arXiv publishes its own query interface, but it returns an Atom feed that has to be parsed and reshaped, and it asks callers to rate-limit themselves. Search1API exposes arXiv as one of the search services on its Search endpoint, returning the paper title, the arxiv.org link, and a snippet carrying the author list, the submission date, and the abstract, in the same JSON shape it returns for its other engines. For research agents this ordering matters: the abstract arrives with the search result, so relevance can be judged before a full-text fetch is spent.
Typical workflow
Search arXiv over HTTP and receive the authors, the submission date, and the abstract in a single field, without parsing the Atom feed the native interface returns.
Send a query with `search_service` set to `arxiv`.
Split `snippet` on the pipe to separate authors, date, and abstract.
Set `crawl_results` when the workflow needs the paper page rather than the abstract.
One request, papers with citation fields
A single call searches arXiv and returns the authors, submission date, and abstract with each result, which is enough to build a citation and judge relevance in one pass.
bash
Best for
Literature reviews that need abstracts before committing to full texts.
Research agents that cite papers with authors and dates.
Trend tracking across a field by submission date.
RAG pipelines that index abstracts as a first retrieval layer.
FAQ
How is this different from the arXiv API?
arXiv’s own interface returns an Atom feed that has to be parsed and reshaped, and it asks callers to throttle themselves. Search1API returns JSON in the same shape as its other engines, with the authors, date, and abstract already assembled into the snippet.
Can I get the full paper text?
Set `crawl_results` to the number of top results you want fetched, and each page retrieved successfully includes a `content` field with the page text as markdown. Results that cannot be retrieved are returned without `content` and are not charged.
Are the results only preprints?
Results come from arXiv, which hosts preprints alongside versions of published work that authors have deposited. The submission date in the snippet is the arXiv date, not a journal publication date.
What does an arXiv search cost?
One credit per search, and one additional credit for each page retrieved successfully through `crawl_results`. At the base top-up rate of $1 for 1,000 credits, that is $1 per 1,000 searches. Signup includes 100 free credits with no credit card.
Can I restrict results to a date range?
Use `time_range` for day, week, month, or year windows. For a specific range, the submission date is returned in the snippet, so results can be filtered after the fact.
Which other research sources are supported?
The same Search endpoint accepts `wikipedia` for reference articles, `github` for implementations, and `reddit` for discussion, alongside general web engines and the Chinese-language engines, for 17 engines in total on one contract.