AI Overview for SearXNG
Source-linked AI summaries alongside your search results, with a choice of local models or hosted providers.
- Python
- JavaScript
- SearXNG
An overview without leaving search
End a search with a question mark to request a summary from the returned snippets. The answer streams above the ordinary results, with numbered citations that open its source snippets. You can stop, retry, or regenerate the answer, and optionally choose from configured models.
Choose where generation runs
Generation can run locally through Ollama or use a hosted provider. Provider credentials stay on the server, and numbered citations make the snippets behind an answer easy to inspect.
One interface, several providers
A Python service normalizes provider streams for a JavaScript interface that handles cancellation, retries, and rendering. The browser modules ship with the plugin without a frontend build step. A standalone demo uses recorded answers, making the interface testable without a model server. The project currently targets personal SearXNG instances using the Simple theme.