row_type,pillar,block,metric,label,chatgpt_value,claude_value,gemini_value,chatgpt_score_0_10,claude_score_0_10,gemini_score_0_10,scoring_rule,source_file
metric,capability,arena_overall_days,arena_overall_days,"Arena text overall: share of days at #1, last 18 months",0.135,0.438,0.427,3.08,10.0,9.75,10 x value / best,lmarena_days_at_1.csv
metric,capability,arena_overall_now,arena_overall_now,Arena text overall: current best rating (2026-10-22),1485.14,1508.73,1517.17,9.08,9.76,10.0,10 x Elo win-probability vs leader / 0.5,lmarena_history.csv
metric,capability,eci,eci,"Epoch Capabilities Index, best current model",168.81,168.83,166.45,10.0,10.0,10.0,"tie if 95% CI overlaps leader, else min-max",raw/epoch/.../eci_scores.csv
metric,capability,independent_benchmarks,GPQA diamond,GPQA diamond (Epoch AI own run),0.9577,0.9558,0.964,9.93,9.91,10.0,10 x value / best,benchmarks_current.csv
metric,capability,independent_benchmarks,SimpleQA Verified,SimpleQA Verified (Epoch AI own run),0.756,0.722,0.76,9.95,9.5,10.0,10 x value / best,benchmarks_current.csv
metric,capability,independent_benchmarks,FrontierMath-Tiers-1-3-v2-Private,FrontierMath-Tiers-1-3-v2-Private (Epoch AI own run),0.9368,0.9123,0.8596,10.0,9.74,9.18,10 x value / best,benchmarks_current.csv
metric,capability,independent_benchmarks,FrontierMath-Tier-4-v2-Private,FrontierMath-Tier-4-v2-Private (Epoch AI own run),1.0,0.95,0.7805,10.0,9.5,7.8,10 x value / best,benchmarks_current.csv
metric,capability,independent_benchmarks,Chess Puzzles,Chess Puzzles (Epoch AI own run),0.76,0.67,0.59,10.0,8.82,7.76,10 x value / best,benchmarks_current.csv
metric,capability,independent_benchmarks,Mystery Game Puzzles,Mystery Game Puzzles (Epoch AI own run),0.84,0.71,0.64,10.0,8.45,7.62,10 x value / best,benchmarks_current.csv
metric,capability,independent_benchmarks,Furniture Assembly,Furniture Assembly (Epoch AI own run),0.8,0.8333,0.725,9.6,10.0,8.7,10 x value / best,benchmarks_current.csv
metric,capability,independent_benchmarks,EBR-bench,EBR-bench (Epoch AI own run),0.7619,0.7143,0.6476,10.0,9.38,8.5,10 x value / best,benchmarks_current.csv
metric,coding,coding_track_days,arena_coding_days,"Arena coding: share of days at #1, last 18 months",0.122,0.746,0.126,1.64,10.0,1.69,10 x value / best,lmarena_days_at_1.csv
metric,coding,coding_track_days,webdev_days,Code Arena (WebDev): share of days at #1 since Nov 2025,0.063,0.873,0.029,0.72,10.0,0.33,10 x value / best,lmarena_days_at_1.csv
metric,coding,coding_votes_now,arena_coding_now,Arena coding: current best rating,1543.07,1550.56,1551.57,9.76,9.97,10.0,10 x Elo win-probability vs leader / 0.5,lmarena_history.csv
metric,coding,coding_votes_now,webdev_now,Code Arena (WebDev): current best rating,1774.31,1806.92,1674.18,9.06,10.0,6.36,10 x Elo win-probability vs leader / 0.5,lmarena_history.csv
metric,coding,coding_tests,MirrorCode,MirrorCode (Epoch AI own run),0.467,0.774,0.5889,6.03,10.0,7.61,10 x value / best,benchmarks_current.csv
metric,coding,coding_tests,FrontierSWE,FrontierSWE (Epoch-compiled third-party leaderboard),0.6551,0.6233,0.5496,10.0,9.51,8.39,10 x value / best,benchmarks_current.csv
metric,coding,coding_tests,WeirdML v3,WeirdML v3 (Epoch-compiled third-party leaderboard),0.5405,0.4983,0.4136,10.0,9.22,7.65,10 x value / best,benchmarks_current.csv
metric,coding,coding_tests,FrontierCode,FrontierCode (Epoch-compiled third-party leaderboard),0.5326,0.5464,0.4755,9.75,10.0,8.7,10 x value / best,benchmarks_current.csv
metric,reliability,app_days_mc,app_days_mc,Consumer app: days with a major/critical incident (of 572),82.0,168.0,,10.0,4.88,,10 x best / value,reliability_summary.csv
metric,reliability,api_days_mc,api_days_mc,API: days with a major/critical incident (of 572),41.0,157.0,,10.0,2.61,,10 x best / value,reliability_summary.csv
metric,reliability,hours_mc,hours_mc,All products: hours with a major/critical incident open,501.9,1397.3,,10.0,3.59,,10 x best / value,reliability_summary.csv
metric,reliability,median_dur,median_dur,Median incident duration (hours),1.65,1.1,,6.67,10.0,,10 x best / value,reliability_summary.csv
metric,value,cheapest_paid,cheapest_paid,Cheapest paid consumer plan ($/mo),8.0,20.0,4.99,6.24,2.5,10.0,10 x best / value,consumer_plans.csv
metric,value,tier20,tier20,"~$20 plan, lowest verified monthly-equivalent ($)",20.0,17.0,19.99,8.5,10.0,8.5,10 x best / value,consumer_plans.csv
metric,value,api_eci166,api_eci166,API: cheapest model with ECI >= 166 (blended $/M),4.0,4.0,4.0,10.0,10.0,10.0,10 x best / value,prices_big3.csv + ECI
metric,value,api_eci155,api_eci155,API: cheapest model with ECI >= 155 (blended $/M),0.2,0.2,1.5,10.0,10.0,1.33,10 x best / value,prices_big3.csv + ECI
metric,features,paid20,paid20,Feature checklist at ~$20 tier (points of 12),9.256,9.0,11.0,8.41,8.18,10.0,10 x value / best,features_matrix.csv
metric,features,free,free,Free-plan checklist (points of 6),3.057,2.5,3.0,10.0,8.18,9.81,10 x value / best,features_matrix.csv
pillar_score_0_10,capability,,capability_pillar,capability pillar (weight 30%),,,,8.02,9.79,9.61,,
pillar_score_0_10,coding,,coding_pillar,coding pillar (weight 25%),,,,6.51,9.89,5.76,,
pillar_score_0_10,reliability,,reliability_pillar,reliability pillar (weight 15%),,,,9.17,5.27,,,
pillar_score_0_10,value,,value_pillar,value pillar (weight 15%),,,,8.68,8.12,7.46,,
pillar_score_0_10,features,,features_pillar,features pillar (weight 15%),,,,9.21,8.18,9.91,,
overall_0_100,overall,,overall,Overall score (0-100),,,,80.9,86.5,81.5,,