Inspect all 4,881 benchmark and cache cases, including the input, reference answer, provenance and model outputs. An additional 204 cases used in supplementary measurements are available in the evidence appendix. Filters affect this appendix only; the paper’s tables remain the full frozen study.
All study calls are retained: measured requests, warmups, cache priming and setup probes. Expand a row for the exact request, API response, tokens, bill and timing. The saved records omit authorization headers. Full records load on demand from the saved evidence files.