Independent search API study
Search APIs,
measured in public.
Compare retrieval quality, cost, and latency across APIs, using authored questions and judge families.
The published comparison
Different questions. Different strengths.
Category scores out of 10. Select a provider to inspect its evidence.
% complete judge panels · of public responses. Withheld cells fall below the category coverage floor.
Full resultsCoverage and publication notes
A finding from this snapshot
The tradeoffs are in the details.
Overall quality
The evidence behind each score.
Equal-weight category mean, 0–10. Incomplete providers remain unranked; a shared tier indicates an unresolved difference.
Open to inspection.
Follow a question, examine the method, or work with the published measurements.
Explore the questions
Individual judge ratings, response timings, and missing evidence.
Open the query explorer 02 / MethodUnderstand the study
The workload, judging process, statistical approach, and limitations.
Read the methodology 03 / DataInspect the measurements
Dated exports and recorded price sources, ready for your own analysis.
Browse the exportsBuilt with checkpointed retrieval, recoverable judging, and reproducible analysis. Explore the offline engineering demo.