{"updated":"2026-10-28","runs":12,"fields":10,"tasks_total":72,"trials_total":784,"overall":[{"ai":"ChatGPT","correct":191,"trials":207,"accuracy":0.92269999999999996465049889593501575291156768798828125,"accuracy_display":92.2999999999999971578290569595992565155029296875,"consistency":1,"consistency_display":100,"seconds_median":21.88329999999999841975295566953718662261962890625,"source_alive":1,"source_alive_display":100,"instruction_follow":0.96109999999999995434762922741356305778026580810546875,"instruction_follow_display":96.099999999999994315658113919198513031005859375,"runs":12},{"ai":"Claude","correct":146,"trials":190,"accuracy":0.7683999999999999719335619374760426580905914306640625,"accuracy_display":76.7999999999999971578290569595992565155029296875,"consistency":0.949500000000000010658141036401502788066864013671875,"consistency_display":95,"seconds_median":19.2727000000000003865352482534945011138916015625,"source_alive":null,"source_alive_display":null,"instruction_follow":0.53749999999999997779553950749686919152736663818359375,"instruction_follow_display":53.7999999999999971578290569595992565155029296875,"runs":12},{"ai":"Copilot","correct":130,"trials":180,"accuracy":0.72219999999999995310417943983338773250579833984375,"accuracy_display":72.2000000000000028421709430404007434844970703125,"consistency":0.87219999999999997530863993233651854097843170166015625,"consistency_display":87.2000000000000028421709430404007434844970703125,"seconds_median":16.629999999999999005240169935859739780426025390625,"source_alive":0.93330000000000001847411112976260483264923095703125,"source_alive_display":93.2999999999999971578290569595992565155029296875,"instruction_follow":0.76759999999999994901855870921281166374683380126953125,"instruction_follow_display":76.7999999999999971578290569595992565155029296875,"runs":11},{"ai":"Gemini","correct":127,"trials":207,"accuracy":0.61350000000000004529709940470638684928417205810546875,"accuracy_display":61.39999999999999857891452847979962825775146484375,"consistency":0.8606000000000000316191517413244582712650299072265625,"consistency_display":86.099999999999994315658113919198513031005859375,"seconds_median":49.00829999999999841975295566953718662261962890625,"source_alive":0.83330000000000004067857162226573564112186431884765625,"source_alive_display":83.2999999999999971578290569595992565155029296875,"instruction_follow":0.8306999999999999939603867460391484200954437255859375,"instruction_follow_display":83.099999999999994315658113919198513031005859375,"runs":12}],"by_field":[{"field":"Meeting minutes and summaries","top":"ChatGPT","accuracy":0.91669999999999995932142837773426435887813568115234375,"accuracy_display":91.7000000000000028421709430404007434844970703125,"runs":2},{"field":"Reading images and photos","top":"ChatGPT","accuracy":0.8667000000000000259348098552436567842960357666015625,"accuracy_display":86.7000000000000028421709430404007434844970703125,"runs":2},{"field":"Reading PDFs and financial statements","top":"ChatGPT","accuracy":1,"accuracy_display":100,"runs":1},{"field":"Research and fact-checking","top":"ChatGPT","accuracy":1,"accuracy_display":100,"runs":1},{"field":"English text and translation","top":"ChatGPT","accuracy":0.83330000000000004067857162226573564112186431884765625,"accuracy_display":83.2999999999999971578290569595992565155029296875,"runs":1},{"field":"Spreadsheets and aggregation","top":"ChatGPT","accuracy":1,"accuracy_display":100,"runs":1},{"field":"Document outlining","top":"ChatGPT","accuracy":0.83330000000000004067857162226573564112186431884765625,"accuracy_display":83.2999999999999971578290569595992565155029296875,"runs":1},{"field":"Excel and spreadsheet formulas","top":"ChatGPT","accuracy":1,"accuracy_display":100,"runs":1},{"field":"Email and business documents","top":"ChatGPT","accuracy":1,"accuracy_display":100,"runs":1},{"field":"Data cleanup and name matching","top":"ChatGPT","accuracy":0.83330000000000004067857162226573564112186431884765625,"accuracy_display":83.2999999999999971578290569595992565155029296875,"runs":1}],"history":[{"run_date":"2026-10-29","slug":"gijiroku-2026-10-29","field":"Meeting minutes and summaries","winner":"ChatGPT"},{"run_date":"2026-10-24","slug":"gazou-2026-10-24","field":"Reading images and photos","winner":"ChatGPT"},{"run_date":"2026-10-19","slug":"meibo-2026-10-19","field":"Data cleanup and name matching","winner":"ChatGPT"},{"run_date":"2026-10-14","slug":"mail-2026-10-14","field":"Email and business documents","winner":null},{"run_date":"2026-10-08","slug":"suushiki-2026-10-08","field":"Excel and spreadsheet formulas","winner":null},{"run_date":"2026-10-03","slug":"kossi-2026-10-03","field":"Document outlining","winner":"ChatGPT"},{"run_date":"2026-10-02","slug":"hyokeisan-2026-10-02","field":"Spreadsheets and aggregation","winner":null},{"run_date":"2026-09-30","slug":"eibun-2026-09-30","field":"English text and translation","winner":"ChatGPT"},{"run_date":"2026-09-29","slug":"shirabemono-2026-09-29","field":"Research and fact-checking","winner":"ChatGPT"},{"run_date":"2026-09-28","slug":"kessan-2026-09-28","field":"Reading PDFs and financial statements","winner":null},{"run_date":"2026-09-27","slug":"gazou-2026-09-27","field":"Reading images and photos","winner":null},{"run_date":"2026-09-24","slug":"gijiroku-2026-09-24","field":"Meeting minutes and summaries","winner":null}],"method_note":"Even though the daily task differs, the five metrics stay the same. Scores are cumulative totals added up across days. Machine scoring only.","license":"CC BY 4.0","attribution":"Please cite AInformation as the source, along with the URL of the page you used and the measurement date."}