Model::search() that understands meaning

Scout, exactly as documented

Four lines of .env, and every Searchable model gets semantic search. No Solr configuration, no schema design, no embedding model.

SCOUT_DRIVER=opensolr
OPENSOLR_EMAIL=you@example.com
OPENSOLR_API_KEY=your-api-key
OPENSOLR_INDEX=myapp__dense

Post::search('budget dining')
    ->where('category', 'restaurants')
    ->paginate(15);

// hybrid BM25 + kNN, embedded
// server-side, real totals

The whole thing, as an open-source Laravel app

A complete application built on this driver, MIT-licensed, on GitHub and Packagist. Start a project from it with one command, or clone it and read it.

laravel-opensolr-search is a Laravel 13 application with a Vue 3 front end on Vite and exactly one search dependency: opensolr/laravel-scout-opensolr. No Solr client to configure, no schema, no embedding model. The API key stays on the server; the browser only talks to the app's own JSON endpoints.

• Eloquent search through Scout. An Article model with the Searchable trait, twenty seeded articles, Article::search($q)->where('category', $c)->paginate(). “Sleepy pets” finds the piece about cats napping; the category becomes a Solr filter.
• Search over the whole index. One call to embed_and_search: hybrid ranking, highlighting, spelling suggestions, facet counts, a “fresh results first” toggle, and the Search Tuning saved for the index applied automatically.
• A grounded AI answer. An Ask AI button calls aiAnswer(): the platform retrieves the top hits for the question and writes an answer from them, shown next to its sources.
• Built to be copied. Form request validation on every input, per-address rate limits, a Content Security Policy, highlights rendered as text segments rather than HTML, result links validated server-side, and a CI that runs the Composer and npm audits on every push.

It runs on the demo account above, out of the box: the index search reads the shared news index, and one artisan command creates a private index for your models.

composer create-project opensolr/laravel-opensolr-search my-search
cd my-search
php artisan opensolr:create-index
php artisan scout:import "App\Models\Article"
npm install && npm run build && php artisan serve

Search with a photo

Read a picture into words, then run those words through the ordinary search. Opensolr reads the image three ways — what it depicts, the text printed on it, and any barcode or QR code — and nothing new is stored in Solr; the picture simply becomes a query.

use Opensolr\ScoutOpensolr\OpensolrClient;

$client = new OpensolrClient(config('scout-opensolr.email'), config('scout-opensolr.api_key'));

// What the picture reads as — labels, OCR text, barcodes. Accepts a path or raw bytes:
$read = $client->imageToWords('catalog__dense', storage_path('app/shelf-photo.jpg'));
// ['text' => 'red running shoe', 'mode' => 'clip',
//  'labels' => ['running shoe','sneaker'], 'codes' => ['0123456789012']]

// Feed the reading you want into the normal hybrid search:
$body = $client->embedAndSearch('catalog__dense', $read['text'], 5);
$docs = $body['response']['docs'];

Pick which reading drives the search by choosing the field — text (the engine's choice), labels (what it depicts), or codes (an exact barcode match). imageToText() returns the raw API response when you also want the thumbnails.

Tested on our own production index

Every release runs a live end-to-end suite against real Opensolr infrastructure — no mocks.

• Full write path through the async Data Ingestion queue: queued → server-side embeddings & enrichment → searchable, with metadata and id round-trips verified. Metadata lands in typed Solr fields; the suffixes and what they enable are in the Vector Search Schema Reference.
• Real-corpus retrieval against a replica of opensolr.com's own search index: pure-semantic hits with zero keyword overlap („how do I get my data back after a disaster” → backup docs), cross-lingual queries (Romanian → English content), all hybrid modes and the full alpha range.
• PDF ingestion via rtf:true: server-side text extraction and content-type detection, then semantic retrieval from the extracted content.
• Grounded RAG answers: one call runs hybrid retrieval and feeds the top hits to the LLM (configurable docs/words, custom instruction) — verified with a question answerable only from the ingested PDF.
• Scout specifics: real Eloquent models via Orchestra Testbench — searchable() through the queue, where() filters, pagination with real totals, unsearchable() removal.