ORBIT searches about 250 million papers across open scholarly indexes and tells you, for every result, whether there is a legal free copy and where. Searching, filtering and exporting are free and cost no credit. One purchased credit has ORBIT read the results against your own abstract rather than against your keywords.
Open scholarly indexes, read through their own APIs. ORBIT does not crawl publishers’ pages to build this, and the rule behind that is worth stating plainly: an API that already exists is a better source than a crawl of the page in front of it — better data, obtained faster, without reopening the question of what we are entitled to redistribute.
Open-access links come from the same record. Where a paper is free to read, ORBIT links to the legal copy and names the licence it is published under; where it is paywalled, ORBIT says so and links to the DOI. It will not route anyone around a paywall.
Searching, filtering by year or open access, following the links, and exporting to BibTeX or CSV are all free and spend no credit. One search costs ORBIT roughly a hundredth of a US cent, so charging for it would be margin dressed up as cost recovery rather than a price.
One purchased credit buys the part that costs real money: a model reading the retrieved abstracts against yours. It returns three things at once, because they are one reading of the same papers — a relevance score for each result with a reason, what the literature already says about the claim you are making as new, and a draft related-work paragraph.
Every other ORBIT output restates something a source already stated. A paragraph summarising twenty abstracts cannot: it has to compose sentences the author did not write. Rather than blur that, ORBIT states it and enforces the limit in code instead of promising it in prose.
Every sentence in the draft must cite at least one of the papers you were shown, by number. Any sentence that cites nothing, or cites a paper that was not in your results, is deleted before it reaches you — and you are told how many were deleted. It is a draft to check, not a paragraph to paste.
Yes, and there is no cap you will meet by using it normally. A search costs ORBIT about a hundredth of a US cent, so charging for it would be margin dressed up as cost recovery. You spend a credit only if you ask ORBIT to read the results against your own abstract — that is a model call, and that is what actually costs us money.
Open scholarly indexes covering about 250 million papers, read through their own APIs, along with the open-access records that say which of those papers are legally free. ORBIT does not crawl publishers' pages to build this: an API that already exists is a better source than a crawl of the page in front of it — better data, faster, and without reopening the question of whether we are entitled to redistribute it.
Where the paper is open access, yes — ORBIT links straight to the legal free copy and shows you the licence it is published under. Where it is paywalled, ORBIT says so and links to the DOI. It will not route you around a paywall.
Three things for one purchased credit, because they are one reading of the same papers: it scores each result for relevance to your specific abstract and says why; it flags what the literature already says about the claim you are making as new; and it drafts a related-work paragraph. Search results are ranked by keyword — this is ranked by what your paper is about.
This is the one exception, and it is named rather than hidden. Every other ORBIT output restates something a source already stated; a paragraph summarising twenty abstracts has to compose prose. So the rule is enforced instead of promised: every sentence must cite at least one of the papers you were shown, by number, and any sentence that cites nothing is deleted before it reaches you — you are told how many were deleted. Check every citation before you use it. It is a draft, not a finished paragraph.
No. Searches are cached so that two people looking for the same topic share one answer, but the cache holds no user id and cannot be traced back to anyone — the key is a hash of the search itself. There is deliberately no search history: what a researcher is reading is a map of work they have not published yet, and ORBIT has no use for it.
Yes, in four formats and all of them free: RIS for Zotero, Mendeley, EndNote or RefWorks; BibTeX for LaTeX; CSV for a spreadsheet; and a plain Markdown reading list. Fields the record does not carry are left out rather than filled with blanks, so a citation that is missing something is visible the moment you compile rather than silently wrong.