Everything your site search does, by API
Every part of Opensolr site search can be called from your own code: start the crawl of your site, push your own records, search by words and by meaning, get answers written from your pages, read pictures and documents, and look up facts for a chat bot. Your site, your app or your AI agent uses the same account and the same key.
Each area has its own page. Every endpoint is one sentence there, with a link to its full reference: parameters, answers and examples.
Two addresses
The calls that use the AI models and your pushed data go to https://api.opensolr.com. Everything else goes to https://opensolr.com. A call sent to the wrong one answers "not found" and names the right address.
The areas
Start, pause and stop the crawl of your site, have a changed page read again, and watch it live.
Send your records and files, then follow every job until it is in the index.
Results ranked by words and meaning in one call, the fields you can filter by, pinned results.
Answers written from your own pages, the chat bot in one call, a chat model for your agents.
The vectors behind search by meaning, plus language and sentiment, for your own code.
What a picture shows, the text printed in it, documents to text, alt text.
IP and place lookups, local time, exchange and VAT rates, VAT numbers, distances, postal codes.
An index ready for site search by code, then LangChain, LlamaIndex, Haystack, MCP or Laravel Scout.
What every call needs
- Your account email and your API key, or a scoped key that only opens chosen endpoints and indexes. Four calls made for CMS modules are also signed with your key. See API Reference.
- Each account can send 120 requests a minute and 1,200 an hour.
- Calls that use the AI models (vectors, answers, pictures) count against the monthly AI allowance of the account that owns the index. Every plan includes 500 AI requests a month; bigger allowances are on Pricing. A result served from cache costs nothing. What counts, call by call: API Quota.
To see these calls at work on your own content first, try RAG in 60 seconds.