Skip to main content
Google Light Search has a dedicated endpoint: GET https://scrapebadger.com/v1/google/light/search. It returns search_information, organic_results, related_searches and pagination. It skips ads, AI Overviews, Knowledge Graph, People Also Ask, local results, news, videos and shopping blocks. Failed requests are not charged. The response header X-Credits-Used reports the actual charge. Check https://scrapebadger.com/api/public/pricing for current prices and promotions.
No mode parameter is required. Light always uses the lightweight path; mode=full and ai_overview=true cannot enable full Search features. Use start=10 for the next page. num requests 1–100 results; Google may return fewer. Google page variants can have different rankings, even for the same query and locale. search_information.organic_results_state is Results for exact spelling, Results for spelling correction (Google showed results for a corrected spelling), Results with suggested correction (Google suggested a spelling with “Did you mean” but kept your query) or no_results. total_results and time_taken are filled only when Google prints them. Light usually gets Google’s basic page, which shows no result count or search time, so both are often null on Light. Use regular Search when you need them.
Light attempts to resolve every organic result URL within one short lookup budget and skips Google’s verification timers. Unresolved links remain google.com/goto redirects. All results and ranks are retained. See redirect-link behavior. The MCP tool is google_light_search. Both the CLI and MCP call the dedicated Light endpoint. Existing /search?mode=fast clients still work, but that is the regular Search route and retains its 20-credit price. Migrate to /light/search for the 10-credit rate. See the endpoint reference.