{"run_id":"run-20260907T181839-5742df","started_at":1788805119,"summary":{"run_id":"run-20260907T181839-5742df","question_bank_version":"2026-05-17-v1","code_version":"0.2.8","started_at":1788805119.8764877,"judges_active":["gpt-5","claude-opus-4-8","claude-opus-4-7","mistral-large-2"],"by_model_domain":[{"model_id":"claude-fable-5-1","domain":"jura","score_mean":0.869124149659864,"score_ci_lo":0.8297817460317459,"score_ci_hi":0.9055102040816326,"top100":87,"n_cells":35,"closed_accuracy":null,"open_rubric_score":0.869124149659864,"hallu_rate_mean":1.4928571428571429,"hallu_prevalence":0.9428571428571428,"hallu_top100":62,"inter_rater_agreement":0.8020408163265307,"n_judges":4,"score_mean_ex_family":0.8552267573696145,"self_judge_delta":0.013897392290249444,"n_cells_ex_family":35,"false_claim_types":{"az":9,"paragraph":127,"study":0,"number":6,"other":31},"tokens_in_total":22845,"tokens_out_total":98779,"model_cost_usd":5.1674},{"model_id":"claude-fable-5-1","domain":"medizin","score_mean":0.9011069340016709,"score_ci_lo":0.8502245196324144,"score_ci_hi":0.94702903091061,"top100":90,"n_cells":19,"closed_accuracy":null,"open_rubric_score":0.9011069340016709,"hallu_rate_mean":1.1973684210526316,"hallu_prevalence":1,"hallu_top100":70,"inter_rater_agreement":0.8354218880534671,"n_judges":4,"score_mean_ex_family":0.8530179615705931,"self_judge_delta":0.04808897243107779,"n_cells_ex_family":19,"false_claim_types":{"az":0,"paragraph":0,"study":0,"number":23,"other":47},"tokens_in_total":10930,"tokens_out_total":43860,"model_cost_usd":2.3023},{"model_id":"claude-fable-5-1","domain":"steuerrecht","score_mean":0.7534060846560846,"score_ci_lo":0.6567923280423281,"score_ci_hi":0.844563492063492,"top100":75,"n_cells":15,"closed_accuracy":null,"open_rubric_score":0.7534060846560846,"hallu_rate_mean":3.85,"hallu_prevalence":1,"hallu_top100":3,"inter_rater_agreement":0.6942063492063492,"n_judges":4,"score_mean_ex_family":0.7418650793650793,"self_judge_delta":0.011541005291005324,"n_cells_ex_family":15,"false_claim_types":{"az":3,"paragraph":121,"study":0,"number":18,"other":30},"tokens_in_total":9765,"tokens_out_total":95115,"model_cost_usd":4.8534},{"model_id":"claude-fable-5-1","domain":"wirtschaftsrecht","score_mean":0.7280685550082102,"score_ci_lo":0.6264573070607553,"score_ci_hi":0.8111145320197044,"top100":73,"n_cells":29,"closed_accuracy":null,"open_rubric_score":0.7280685550082102,"hallu_rate_mean":3.060344827586207,"hallu_prevalence":1,"hallu_top100":23,"inter_rater_agreement":0.6750136836343733,"n_judges":4,"score_mean_ex_family":0.7108511220580186,"self_judge_delta":0.017217432950191625,"n_cells_ex_family":29,"false_claim_types":{"az":3,"paragraph":193,"study":0,"number":7,"other":61},"tokens_in_total":21571,"tokens_out_total":138816,"model_cost_usd":7.15651},{"model_id":"claude-opus-4-7","domain":"jura","score_mean":0.806267006802721,"score_ci_lo":0.7450198412698412,"score_ci_hi":0.8597959183673469,"top100":81,"n_cells":35,"closed_accuracy":null,"open_rubric_score":0.806267006802721,"hallu_rate_mean":1.2785714285714285,"hallu_prevalence":0.8285714285714286,"hallu_top100":68,"inter_rater_agreement":0.7802040816326531,"n_judges":4,"score_mean_ex_family":0.7990929705215419,"self_judge_delta":0.007174036281179075,"n_cells_ex_family":35,"false_claim_types":{"az":4,"paragraph":78,"study":0,"number":9,"other":32},"tokens_in_total":22950,"tokens_out_total":30278,"model_cost_usd":0.8717},{"model_id":"claude-opus-4-7","domain":"medizin","score_mean":0.8501736111111111,"score_ci_lo":0.7875248015873015,"score_ci_hi":0.9105654761904762,"top100":85,"n_cells":20,"closed_accuracy":null,"open_rubric_score":0.8501736111111111,"hallu_rate_mean":1.025,"hallu_prevalence":0.95,"hallu_top100":74,"inter_rater_agreement":0.8371031746031745,"n_judges":4,"score_mean_ex_family":0.8126984126984127,"self_judge_delta":0.03747519841269842,"n_cells_ex_family":20,"false_claim_types":{"az":0,"paragraph":0,"study":0,"number":17,"other":46},"tokens_in_total":11495,"tokens_out_total":20325,"model_cost_usd":0.5656},{"model_id":"claude-opus-4-7","domain":"steuerrecht","score_mean":0.7296097883597883,"score_ci_lo":0.6575000000000001,"score_ci_hi":0.8012499999999999,"top100":73,"n_cells":15,"closed_accuracy":null,"open_rubric_score":0.7296097883597883,"hallu_rate_mean":2.3833333333333333,"hallu_prevalence":1,"hallu_top100":40,"inter_rater_agreement":0.8084656084656084,"n_judges":4,"score_mean_ex_family":0.7272619047619048,"self_judge_delta":0.0023478835978835155,"n_cells_ex_family":15,"false_claim_types":{"az":4,"paragraph":62,"study":0,"number":18,"other":18},"tokens_in_total":9810,"tokens_out_total":19502,"model_cost_usd":0.5366},{"model_id":"claude-opus-4-7","domain":"wirtschaftsrecht","score_mean":0.7914583333333333,"score_ci_lo":0.7504761904761904,"score_ci_hi":0.8302480158730159,"top100":79,"n_cells":30,"closed_accuracy":null,"open_rubric_score":0.7914583333333333,"hallu_rate_mean":2.566666666666667,"hallu_prevalence":1,"hallu_top100":35,"inter_rater_agreement":0.7139682539682539,"n_judges":4,"score_mean_ex_family":0.7681216931216931,"self_judge_delta":0.02333664021164017,"n_cells_ex_family":30,"false_claim_types":{"az":9,"paragraph":197,"study":0,"number":21,"other":36},"tokens_in_total":22262,"tokens_out_total":42676,"model_cost_usd":1.17821},{"model_id":"claude-opus-4-8","domain":"jura","score_mean":0.8126558956916099,"score_ci_lo":0.7687074829931972,"score_ci_hi":0.8503543083900227,"top100":81,"n_cells":35,"closed_accuracy":null,"open_rubric_score":0.8126558956916099,"hallu_rate_mean":0.7071428571428572,"hallu_prevalence":0.7428571428571429,"hallu_top100":82,"inter_rater_agreement":0.8013151927437642,"n_judges":4,"score_mean_ex_family":0.8218707482993197,"self_judge_delta":-0.009214852607709823,"n_cells_ex_family":35,"false_claim_types":{"az":0,"paragraph":44,"study":0,"number":2,"other":30},"tokens_in_total":22775,"tokens_out_total":31422,"model_cost_usd":0.899425},{"model_id":"claude-opus-4-8","domain":"medizin","score_mean":0.8073660714285713,"score_ci_lo":0.7522569444444445,"score_ci_hi":0.865451388888889,"top100":81,"n_cells":20,"closed_accuracy":null,"open_rubric_score":0.8073660714285713,"hallu_rate_mean":0.775,"hallu_prevalence":0.95,"hallu_top100":80,"inter_rater_agreement":0.7100198412698413,"n_judges":4,"score_mean_ex_family":0.7554563492063492,"self_judge_delta":0.05190972222222212,"n_cells_ex_family":20,"false_claim_types":{"az":0,"paragraph":0,"study":0,"number":15,"other":36},"tokens_in_total":11395,"tokens_out_total":20602,"model_cost_usd":0.572025},{"model_id":"claude-opus-4-8","domain":"steuerrecht","score_mean":0.6880753968253969,"score_ci_lo":0.6350925925925925,"score_ci_hi":0.7394775132275132,"top100":69,"n_cells":15,"closed_accuracy":null,"open_rubric_score":0.6880753968253969,"hallu_rate_mean":1.5333333333333334,"hallu_prevalence":0.9333333333333333,"hallu_top100":61,"inter_rater_agreement":0.7723544973544973,"n_judges":4,"score_mean_ex_family":0.7005687830687831,"self_judge_delta":-0.01249338624338625,"n_cells_ex_family":15,"false_claim_types":{"az":2,"paragraph":51,"study":0,"number":8,"other":16},"tokens_in_total":9735,"tokens_out_total":20402,"model_cost_usd":0.558725},{"model_id":"claude-opus-4-8","domain":"wirtschaftsrecht","score_mean":0.6614616402116401,"score_ci_lo":0.6154993386243386,"score_ci_hi":0.7110615079365079,"top100":66,"n_cells":30,"closed_accuracy":null,"open_rubric_score":0.6614616402116401,"hallu_rate_mean":1.9,"hallu_prevalence":0.9666666666666667,"hallu_top100":52,"inter_rater_agreement":0.6344444444444444,"n_judges":4,"score_mean_ex_family":0.5859259259259259,"self_judge_delta":0.07553571428571426,"n_cells_ex_family":30,"false_claim_types":{"az":1,"paragraph":115,"study":0,"number":11,"other":36},"tokens_in_total":22112,"tokens_out_total":45059,"model_cost_usd":1.237035},{"model_id":"claude-opus-5","domain":"jura","score_mean":0.8988633786848073,"score_ci_lo":0.8386734693877551,"score_ci_hi":0.9426332199546485,"top100":90,"n_cells":35,"closed_accuracy":null,"open_rubric_score":0.8988633786848073,"hallu_rate_mean":1.6928571428571428,"hallu_prevalence":0.9428571428571428,"hallu_top100":57,"inter_rater_agreement":0.7837414965986395,"n_judges":4,"score_mean_ex_family":0.9005612244897959,"self_judge_delta":-0.0016978458049886624,"n_cells_ex_family":35,"false_claim_types":{"az":4,"paragraph":143,"study":0,"number":17,"other":36},"tokens_in_total":22775,"tokens_out_total":125745,"model_cost_usd":3.2575},{"model_id":"claude-opus-5","domain":"medizin","score_mean":0.8969990079365079,"score_ci_lo":0.8291170634920635,"score_ci_hi":0.9546875,"top100":90,"n_cells":20,"closed_accuracy":null,"open_rubric_score":0.8969990079365079,"hallu_rate_mean":1.65,"hallu_prevalence":1,"hallu_top100":58,"inter_rater_agreement":0.8706349206349205,"n_judges":4,"score_mean_ex_family":0.8624007936507937,"self_judge_delta":0.03459821428571419,"n_cells_ex_family":20,"false_claim_types":{"az":0,"paragraph":0,"study":0,"number":35,"other":64},"tokens_in_total":11395,"tokens_out_total":68234,"model_cost_usd":1.762825},{"model_id":"claude-opus-5","domain":"steuerrecht","score_mean":0.8186177248677249,"score_ci_lo":0.7300462962962962,"score_ci_hi":0.8943187830687831,"top100":82,"n_cells":15,"closed_accuracy":null,"open_rubric_score":0.8186177248677249,"hallu_rate_mean":3.4,"hallu_prevalence":1,"hallu_top100":15,"inter_rater_agreement":0.7514021164021164,"n_judges":4,"score_mean_ex_family":0.7734391534391534,"self_judge_delta":0.04517857142857151,"n_cells_ex_family":15,"false_claim_types":{"az":4,"paragraph":104,"study":0,"number":12,"other":39},"tokens_in_total":9735,"tokens_out_total":118870,"model_cost_usd":3.020425},{"model_id":"claude-opus-5","domain":"wirtschaftsrecht","score_mean":0.811656746031746,"score_ci_lo":0.7378373015873015,"score_ci_hi":0.8738690476190476,"top100":81,"n_cells":30,"closed_accuracy":null,"open_rubric_score":0.811656746031746,"hallu_rate_mean":2.9833333333333334,"hallu_prevalence":1,"hallu_top100":25,"inter_rater_agreement":0.7576455026455027,"n_judges":4,"score_mean_ex_family":0.750707671957672,"self_judge_delta":0.06094907407407402,"n_cells_ex_family":30,"false_claim_types":{"az":9,"paragraph":197,"study":0,"number":10,"other":65},"tokens_in_total":22112,"tokens_out_total":216143,"model_cost_usd":5.514135},{"model_id":"deepseek-v4-pro","domain":"jura","score_mean":0.5921343537414966,"score_ci_lo":0.5188492063492063,"score_ci_hi":0.658015873015873,"top100":59,"n_cells":35,"closed_accuracy":null,"open_rubric_score":0.5921343537414966,"hallu_rate_mean":0.9428571428571428,"hallu_prevalence":0.6571428571428571,"hallu_top100":76,"inter_rater_agreement":0.6766780045351474,"n_judges":4,"score_mean_ex_family":0.5921343537414966,"self_judge_delta":0,"n_cells_ex_family":35,"false_claim_types":{"az":0,"paragraph":39,"study":0,"number":3,"other":16},"tokens_in_total":16523,"tokens_out_total":251775,"model_cost_usd":0.226232},{"model_id":"deepseek-v4-pro","domain":"medizin","score_mean":0.6985119047619047,"score_ci_lo":0.6136904761904762,"score_ci_hi":0.7715029761904761,"top100":70,"n_cells":20,"closed_accuracy":null,"open_rubric_score":0.6985119047619047,"hallu_rate_mean":0.575,"hallu_prevalence":0.7,"hallu_top100":85,"inter_rater_agreement":0.7609126984126984,"n_judges":4,"score_mean_ex_family":0.6985119047619047,"self_judge_delta":0,"n_cells_ex_family":20,"false_claim_types":{"az":1,"paragraph":0,"study":1,"number":7,"other":19},"tokens_in_total":8383,"tokens_out_total":46478,"model_cost_usd":0.044082},{"model_id":"deepseek-v4-pro","domain":"steuerrecht","score_mean":0.504499716553288,"score_ci_lo":0.416985544217687,"score_ci_hi":0.5768849206349207,"top100":50,"n_cells":14,"closed_accuracy":null,"open_rubric_score":0.504499716553288,"hallu_rate_mean":2.357142857142857,"hallu_prevalence":0.7857142857142857,"hallu_top100":41,"inter_rater_agreement":0.6223922902494331,"n_judges":4,"score_mean_ex_family":0.504499716553288,"self_judge_delta":0,"n_cells_ex_family":14,"false_claim_types":{"az":0,"paragraph":31,"study":0,"number":36,"other":17},"tokens_in_total":6466,"tokens_out_total":126139,"model_cost_usd":0.112554},{"model_id":"deepseek-v4-pro","domain":"wirtschaftsrecht","score_mean":0.5811268472906405,"score_ci_lo":0.4987787356321839,"score_ci_hi":0.6542692939244664,"top100":58,"n_cells":29,"closed_accuracy":null,"open_rubric_score":0.5811268472906405,"hallu_rate_mean":1.5258620689655173,"hallu_prevalence":0.9655172413793104,"hallu_top100":61,"inter_rater_agreement":0.5577586206896552,"n_judges":4,"score_mean_ex_family":0.5811268472906405,"self_judge_delta":0,"n_cells_ex_family":29,"false_claim_types":{"az":0,"paragraph":60,"study":0,"number":11,"other":33},"tokens_in_total":14785,"tokens_out_total":204625,"model_cost_usd":0.184455},{"model_id":"gemini-3-1-pro","domain":"jura","score_mean":0.627610544217687,"score_ci_lo":0.5571938775510203,"score_ci_hi":0.6906179138321995,"top100":63,"n_cells":35,"closed_accuracy":null,"open_rubric_score":0.627610544217687,"hallu_rate_mean":0.65,"hallu_prevalence":0.5714285714285714,"hallu_top100":83,"inter_rater_agreement":0.7138208616780045,"n_judges":4,"score_mean_ex_family":0.627610544217687,"self_judge_delta":0,"n_cells_ex_family":35,"false_claim_types":{"az":0,"paragraph":15,"study":0,"number":2,"other":29},"tokens_in_total":12898,"tokens_out_total":6587,"model_cost_usd":0.10484},{"model_id":"gemini-3-1-pro","domain":"medizin","score_mean":0.602827380952381,"score_ci_lo":0.5303323412698413,"score_ci_hi":0.6716269841269841,"top100":60,"n_cells":20,"closed_accuracy":null,"open_rubric_score":0.602827380952381,"hallu_rate_mean":0.5,"hallu_prevalence":0.7,"hallu_top100":87,"inter_rater_agreement":0.7484126984126984,"n_judges":4,"score_mean_ex_family":0.602827380952381,"self_judge_delta":0,"n_cells_ex_family":20,"false_claim_types":{"az":0,"paragraph":0,"study":0,"number":2,"other":25},"tokens_in_total":6530,"tokens_out_total":4397,"model_cost_usd":0.065824},{"model_id":"gemini-3-1-pro","domain":"steuerrecht","score_mean":0.4458994708994709,"score_ci_lo":0.3763888888888889,"score_ci_hi":0.5273015873015873,"top100":45,"n_cells":15,"closed_accuracy":null,"open_rubric_score":0.4458994708994709,"hallu_rate_mean":1.3833333333333333,"hallu_prevalence":0.7333333333333333,"hallu_top100":65,"inter_rater_agreement":0.6414021164021164,"n_judges":4,"score_mean_ex_family":0.4458994708994709,"self_judge_delta":0,"n_cells_ex_family":15,"false_claim_types":{"az":0,"paragraph":8,"study":0,"number":11,"other":25},"tokens_in_total":5420,"tokens_out_total":3570,"model_cost_usd":0.05368},{"model_id":"gemini-3-1-pro","domain":"wirtschaftsrecht","score_mean":0.512635582010582,"score_ci_lo":0.4298809523809524,"score_ci_hi":0.5932208994708995,"top100":51,"n_cells":30,"closed_accuracy":null,"open_rubric_score":0.512635582010582,"hallu_rate_mean":1.425,"hallu_prevalence":0.9,"hallu_top100":64,"inter_rater_agreement":0.6064550264550265,"n_judges":4,"score_mean_ex_family":0.512635582010582,"self_judge_delta":0,"n_cells_ex_family":30,"false_claim_types":{"az":0,"paragraph":30,"study":0,"number":5,"other":47},"tokens_in_total":11545,"tokens_out_total":9232,"model_cost_usd":0.133874},{"model_id":"glm-5.3","domain":"jura","score_mean":0.7924107142857143,"score_ci_lo":0.7290426587301587,"score_ci_hi":0.8502728174603175,"top100":79,"n_cells":28,"closed_accuracy":null,"open_rubric_score":0.7924107142857143,"hallu_rate_mean":1.9017857142857142,"hallu_prevalence":0.8928571428571429,"hallu_top100":52,"inter_rater_agreement":0.7622023809523809,"n_judges":4,"score_mean_ex_family":0.7924107142857143,"self_judge_delta":0,"n_cells_ex_family":28,"false_claim_types":{"az":0,"paragraph":98,"study":0,"number":7,"other":36},"tokens_in_total":10521,"tokens_out_total":228699,"model_cost_usd":0.50945},{"model_id":"glm-5.3","domain":"medizin","score_mean":0.8419890873015874,"score_ci_lo":0.7954613095238094,"score_ci_hi":0.8876488095238095,"top100":84,"n_cells":20,"closed_accuracy":null,"open_rubric_score":0.8419890873015874,"hallu_rate_mean":1.0875,"hallu_prevalence":1,"hallu_top100":72,"inter_rater_agreement":0.7284722222222222,"n_judges":4,"score_mean_ex_family":0.8419890873015874,"self_judge_delta":0,"n_cells_ex_family":20,"false_claim_types":{"az":1,"paragraph":0,"study":1,"number":25,"other":41},"tokens_in_total":6938,"tokens_out_total":106934,"model_cost_usd":0.239418},{"model_id":"glm-5.3","domain":"steuerrecht","score_mean":0.46458333333333335,"score_ci_lo":0.184375,"score_ci_hi":0.7447916666666666,"top100":46,"n_cells":4,"closed_accuracy":null,"open_rubric_score":0.46458333333333335,"hallu_rate_mean":4.6875,"hallu_prevalence":1,"hallu_top100":0,"inter_rater_agreement":0.5347222222222222,"n_judges":4,"score_mean_ex_family":0.46458333333333335,"self_judge_delta":0,"n_cells_ex_family":4,"false_claim_types":{"az":4,"paragraph":27,"study":0,"number":5,"other":15},"tokens_in_total":1307,"tokens_out_total":51763,"model_cost_usd":0.114663},{"model_id":"glm-5.3","domain":"wirtschaftsrecht","score_mean":0.7895833333333333,"score_ci_lo":0.6541666666666667,"score_ci_hi":0.89375,"top100":79,"n_cells":8,"closed_accuracy":null,"open_rubric_score":0.7895833333333333,"hallu_rate_mean":2.625,"hallu_prevalence":1,"hallu_top100":34,"inter_rater_agreement":0.7712797619047619,"n_judges":4,"score_mean_ex_family":0.7895833333333333,"self_judge_delta":0,"n_cells_ex_family":8,"false_claim_types":{"az":0,"paragraph":43,"study":0,"number":2,"other":22},"tokens_in_total":2970,"tokens_out_total":98688,"model_cost_usd":0.218896},{"model_id":"gpt-5","domain":"jura","score_mean":0.7503883219954649,"score_ci_lo":0.6919047619047619,"score_ci_hi":0.8039880952380952,"top100":75,"n_cells":35,"closed_accuracy":null,"open_rubric_score":0.7503883219954649,"hallu_rate_mean":1.05,"hallu_prevalence":0.9142857142857143,"hallu_top100":73,"inter_rater_agreement":0.731904761904762,"n_judges":4,"score_mean_ex_family":0.7622071050642479,"self_judge_delta":-0.011818783068783034,"n_cells_ex_family":35,"false_claim_types":{"az":0,"paragraph":69,"study":0,"number":5,"other":30},"tokens_in_total":11850,"tokens_out_total":58199,"model_cost_usd":0.932235},{"model_id":"gpt-5","domain":"medizin","score_mean":0.8505208333333334,"score_ci_lo":0.7869791666666667,"score_ci_hi":0.9006944444444445,"top100":85,"n_cells":20,"closed_accuracy":null,"open_rubric_score":0.8505208333333334,"hallu_rate_mean":0.8875,"hallu_prevalence":0.9,"hallu_top100":77,"inter_rater_agreement":0.7909722222222222,"n_judges":4,"score_mean_ex_family":0.882638888888889,"self_judge_delta":-0.03211805555555558,"n_cells_ex_family":20,"false_claim_types":{"az":0,"paragraph":0,"study":0,"number":15,"other":36},"tokens_in_total":6251,"tokens_out_total":27746,"model_cost_usd":0.447445},{"model_id":"gpt-5","domain":"steuerrecht","score_mean":0.6231547619047618,"score_ci_lo":0.48857142857142855,"score_ci_hi":0.7482010582010582,"top100":62,"n_cells":15,"closed_accuracy":null,"open_rubric_score":0.6231547619047618,"hallu_rate_mean":2.183333333333333,"hallu_prevalence":0.9333333333333333,"hallu_top100":45,"inter_rater_agreement":0.6843650793650794,"n_judges":4,"score_mean_ex_family":0.6564550264550264,"self_judge_delta":-0.03330026455026458,"n_cells_ex_family":15,"false_claim_types":{"az":0,"paragraph":34,"study":0,"number":22,"other":27},"tokens_in_total":5093,"tokens_out_total":40057,"model_cost_usd":0.62632},{"model_id":"gpt-5","domain":"wirtschaftsrecht","score_mean":0.8140740740740741,"score_ci_lo":0.7505621693121693,"score_ci_hi":0.8685416666666667,"top100":81,"n_cells":30,"closed_accuracy":null,"open_rubric_score":0.8140740740740741,"hallu_rate_mean":1.3666666666666667,"hallu_prevalence":0.9666666666666667,"hallu_top100":65,"inter_rater_agreement":0.7944576719576719,"n_judges":4,"score_mean_ex_family":0.8465343915343915,"self_judge_delta":-0.03246031746031741,"n_cells_ex_family":30,"false_claim_types":{"az":0,"paragraph":56,"study":0,"number":25,"other":49},"tokens_in_total":11129,"tokens_out_total":68815,"model_cost_usd":1.08787},{"model_id":"gpt-5.6-sol","domain":"jura","score_mean":0.7527352607709751,"score_ci_lo":0.6868792517006803,"score_ci_hi":0.8122222222222222,"top100":75,"n_cells":35,"closed_accuracy":null,"open_rubric_score":0.7527352607709751,"hallu_rate_mean":1.2214285714285715,"hallu_prevalence":0.8857142857142857,"hallu_top100":69,"inter_rater_agreement":0.7221201814058956,"n_judges":4,"score_mean_ex_family":0.7835071806500379,"self_judge_delta":-0.030771919879062803,"n_cells_ex_family":35,"false_claim_types":{"az":0,"paragraph":71,"study":0,"number":12,"other":32},"tokens_in_total":11850,"tokens_out_total":127502,"model_cost_usd":3.88431},{"model_id":"gpt-5.6-sol","domain":"medizin","score_mean":0.8314732142857142,"score_ci_lo":0.7826140873015873,"score_ci_hi":0.8792906746031746,"top100":83,"n_cells":20,"closed_accuracy":null,"open_rubric_score":0.8314732142857142,"hallu_rate_mean":0.9375,"hallu_prevalence":1,"hallu_top100":76,"inter_rater_agreement":0.7693452380952381,"n_judges":4,"score_mean_ex_family":0.8766534391534391,"self_judge_delta":-0.04518022486772488,"n_cells_ex_family":20,"false_claim_types":{"az":0,"paragraph":0,"study":1,"number":13,"other":39},"tokens_in_total":6251,"tokens_out_total":29620,"model_cost_usd":0.919855},{"model_id":"gpt-5.6-sol","domain":"steuerrecht","score_mean":0.6750330687830688,"score_ci_lo":0.5622222222222223,"score_ci_hi":0.7906944444444445,"top100":68,"n_cells":15,"closed_accuracy":null,"open_rubric_score":0.6750330687830688,"hallu_rate_mean":2.466666666666667,"hallu_prevalence":1,"hallu_top100":38,"inter_rater_agreement":0.7312433862433863,"n_judges":4,"score_mean_ex_family":0.7036684303350971,"self_judge_delta":-0.028635361552028282,"n_cells_ex_family":15,"false_claim_types":{"az":0,"paragraph":36,"study":0,"number":26,"other":27},"tokens_in_total":5093,"tokens_out_total":118221,"model_cost_usd":3.572095},{"model_id":"gpt-5.6-sol","domain":"wirtschaftsrecht","score_mean":0.7637301587301587,"score_ci_lo":0.7005853174603175,"score_ci_hi":0.8272123015873016,"top100":76,"n_cells":30,"closed_accuracy":null,"open_rubric_score":0.7637301587301587,"hallu_rate_mean":1.6083333333333334,"hallu_prevalence":1,"hallu_top100":59,"inter_rater_agreement":0.7431349206349206,"n_judges":4,"score_mean_ex_family":0.8092592592592592,"self_judge_delta":-0.04552910052910053,"n_cells_ex_family":30,"false_claim_types":{"az":0,"paragraph":81,"study":0,"number":14,"other":37},"tokens_in_total":11129,"tokens_out_total":187840,"model_cost_usd":5.690845},{"model_id":"gpt-6-astra","domain":"jura","score_mean":0.7115107709750567,"score_ci_lo":0.6520663265306123,"score_ci_hi":0.7697222222222222,"top100":71,"n_cells":35,"closed_accuracy":null,"open_rubric_score":0.7115107709750567,"hallu_rate_mean":1.2714285714285714,"hallu_prevalence":0.9142857142857143,"hallu_top100":68,"inter_rater_agreement":0.6576757369614512,"n_judges":4,"score_mean_ex_family":0.7537074829931972,"self_judge_delta":-0.04219671201814057,"n_cells_ex_family":35,"false_claim_types":{"az":0,"paragraph":74,"study":0,"number":6,"other":35},"tokens_in_total":11850,"tokens_out_total":107477,"model_cost_usd":5.49235},{"model_id":"gpt-6-astra","domain":"medizin","score_mean":0.8396825396825397,"score_ci_lo":0.7938492063492063,"score_ci_hi":0.8851934523809524,"top100":84,"n_cells":20,"closed_accuracy":null,"open_rubric_score":0.8396825396825397,"hallu_rate_mean":0.875,"hallu_prevalence":0.95,"hallu_top100":78,"inter_rater_agreement":0.8150793650793651,"n_judges":4,"score_mean_ex_family":0.878968253968254,"self_judge_delta":-0.03928571428571426,"n_cells_ex_family":20,"false_claim_types":{"az":0,"paragraph":0,"study":0,"number":18,"other":29},"tokens_in_total":6251,"tokens_out_total":37864,"model_cost_usd":1.95571},{"model_id":"gpt-6-astra","domain":"steuerrecht","score_mean":0.6919510582010582,"score_ci_lo":0.5982142857142858,"score_ci_hi":0.773723544973545,"top100":69,"n_cells":15,"closed_accuracy":null,"open_rubric_score":0.6919510582010582,"hallu_rate_mean":2.45,"hallu_prevalence":1,"hallu_top100":38,"inter_rater_agreement":0.7082804232804233,"n_judges":4,"score_mean_ex_family":0.7103439153439154,"self_judge_delta":-0.01839285714285721,"n_cells_ex_family":15,"false_claim_types":{"az":0,"paragraph":32,"study":0,"number":29,"other":25},"tokens_in_total":5093,"tokens_out_total":92629,"model_cost_usd":4.68238},{"model_id":"gpt-6-astra","domain":"wirtschaftsrecht","score_mean":0.7621560846560846,"score_ci_lo":0.6934755291005291,"score_ci_hi":0.8241170634920636,"top100":76,"n_cells":30,"closed_accuracy":null,"open_rubric_score":0.7621560846560846,"hallu_rate_mean":1.4333333333333333,"hallu_prevalence":0.9666666666666667,"hallu_top100":64,"inter_rater_agreement":0.7319576719576719,"n_judges":4,"score_mean_ex_family":0.8051984126984127,"self_judge_delta":-0.043042328042328104,"n_cells_ex_family":30,"false_claim_types":{"az":0,"paragraph":74,"study":0,"number":11,"other":33},"tokens_in_total":11129,"tokens_out_total":172286,"model_cost_usd":8.72559},{"model_id":"grok-4.5","domain":"jura","score_mean":0.6767006802721088,"score_ci_lo":0.6175878684807256,"score_ci_hi":0.7280753968253968,"top100":68,"n_cells":35,"closed_accuracy":null,"open_rubric_score":0.6767006802721088,"hallu_rate_mean":0.42857142857142855,"hallu_prevalence":0.4857142857142857,"hallu_top100":89,"inter_rater_agreement":0.7242176870748299,"n_judges":4,"score_mean_ex_family":0.6767006802721088,"self_judge_delta":0,"n_cells_ex_family":35,"false_claim_types":{"az":0,"paragraph":8,"study":0,"number":1,"other":15},"tokens_in_total":28906,"tokens_out_total":7070,"model_cost_usd":0.100232},{"model_id":"grok-4.5","domain":"medizin","score_mean":0.6965029761904762,"score_ci_lo":0.6394097222222223,"score_ci_hi":0.7508184523809524,"top100":70,"n_cells":20,"closed_accuracy":null,"open_rubric_score":0.6965029761904762,"hallu_rate_mean":0.2125,"hallu_prevalence":0.5,"hallu_top100":94,"inter_rater_agreement":0.6982142857142857,"n_judges":4,"score_mean_ex_family":0.6965029761904762,"self_judge_delta":0,"n_cells_ex_family":20,"false_claim_types":{"az":0,"paragraph":0,"study":0,"number":2,"other":12},"tokens_in_total":15932,"tokens_out_total":5342,"model_cost_usd":0.063916},{"model_id":"grok-4.5","domain":"steuerrecht","score_mean":0.5520899470899471,"score_ci_lo":0.47347883597883594,"score_ci_hi":0.631547619047619,"top100":55,"n_cells":15,"closed_accuracy":null,"open_rubric_score":0.5520899470899471,"hallu_rate_mean":1.2,"hallu_prevalence":0.7333333333333333,"hallu_top100":70,"inter_rater_agreement":0.6428042328042328,"n_judges":4,"score_mean_ex_family":0.5520899470899471,"self_judge_delta":0,"n_cells_ex_family":15,"false_claim_types":{"az":0,"paragraph":13,"study":0,"number":6,"other":16},"tokens_in_total":12303,"tokens_out_total":5105,"model_cost_usd":0.055236},{"model_id":"grok-4.5","domain":"wirtschaftsrecht","score_mean":0.6847420634920635,"score_ci_lo":0.6348214285714285,"score_ci_hi":0.7349801587301587,"top100":68,"n_cells":30,"closed_accuracy":null,"open_rubric_score":0.6847420634920635,"hallu_rate_mean":0.95,"hallu_prevalence":0.8,"hallu_top100":76,"inter_rater_agreement":0.6306349206349205,"n_judges":4,"score_mean_ex_family":0.6847420634920635,"self_judge_delta":0,"n_cells_ex_family":30,"false_claim_types":{"az":0,"paragraph":35,"study":0,"number":5,"other":39},"tokens_in_total":25515,"tokens_out_total":14602,"model_cost_usd":0.138642},{"model_id":"grok-4.6","domain":"jura","score_mean":0.6344614512471655,"score_ci_lo":0.555890022675737,"score_ci_hi":0.6987301587301588,"top100":63,"n_cells":35,"closed_accuracy":null,"open_rubric_score":0.6344614512471655,"hallu_rate_mean":0.7785714285714286,"hallu_prevalence":0.6285714285714286,"hallu_top100":80,"inter_rater_agreement":0.7025510204081633,"n_judges":4,"score_mean_ex_family":0.6344614512471655,"self_judge_delta":0,"n_cells_ex_family":35,"false_claim_types":{"az":0,"paragraph":19,"study":0,"number":7,"other":25},"tokens_in_total":33876,"tokens_out_total":7284,"model_cost_usd":0.111456},{"model_id":"grok-4.6","domain":"medizin","score_mean":0.6200396825396826,"score_ci_lo":0.540873015873016,"score_ci_hi":0.6998263888888889,"top100":62,"n_cells":20,"closed_accuracy":null,"open_rubric_score":0.6200396825396826,"hallu_rate_mean":0.5,"hallu_prevalence":0.65,"hallu_top100":87,"inter_rater_agreement":0.6841269841269841,"n_judges":4,"score_mean_ex_family":0.6200396825396826,"self_judge_delta":0,"n_cells_ex_family":20,"false_claim_types":{"az":0,"paragraph":0,"study":0,"number":2,"other":23},"tokens_in_total":18772,"tokens_out_total":4467,"model_cost_usd":0.064346},{"model_id":"grok-4.6","domain":"steuerrecht","score_mean":0.599021164021164,"score_ci_lo":0.4993518518518519,"score_ci_hi":0.6948544973544973,"top100":60,"n_cells":15,"closed_accuracy":null,"open_rubric_score":0.599021164021164,"hallu_rate_mean":1.7833333333333334,"hallu_prevalence":0.9333333333333333,"hallu_top100":55,"inter_rater_agreement":0.6965079365079365,"n_judges":4,"score_mean_ex_family":0.599021164021164,"self_judge_delta":0,"n_cells_ex_family":15,"false_claim_types":{"az":0,"paragraph":32,"study":0,"number":7,"other":28},"tokens_in_total":14433,"tokens_out_total":6456,"model_cost_usd":0.067602},{"model_id":"grok-4.6","domain":"wirtschaftsrecht","score_mean":0.639646164021164,"score_ci_lo":0.5616402116402116,"score_ci_hi":0.7115079365079365,"top100":64,"n_cells":30,"closed_accuracy":null,"open_rubric_score":0.639646164021164,"hallu_rate_mean":1.325,"hallu_prevalence":0.9333333333333333,"hallu_top100":66,"inter_rater_agreement":0.6781746031746032,"n_judges":4,"score_mean_ex_family":0.639646164021164,"self_judge_delta":0,"n_cells_ex_family":30,"false_claim_types":{"az":0,"paragraph":53,"study":0,"number":6,"other":36},"tokens_in_total":29775,"tokens_out_total":15326,"model_cost_usd":0.151506},{"model_id":"kimi-k3","domain":"jura","score_mean":0.778315897065897,"score_ci_lo":0.696043771043771,"score_ci_hi":0.8446248196248196,"top100":78,"n_cells":33,"closed_accuracy":null,"open_rubric_score":0.778315897065897,"hallu_rate_mean":1.8863636363636365,"hallu_prevalence":0.9393939393939394,"hallu_top100":52,"inter_rater_agreement":0.8019961519961519,"n_judges":4,"score_mean_ex_family":0.778315897065897,"self_judge_delta":0,"n_cells_ex_family":33,"false_claim_types":{"az":3,"paragraph":118,"study":0,"number":4,"other":33},"tokens_in_total":16805,"tokens_out_total":180748,"model_cost_usd":0.738957},{"model_id":"kimi-k3","domain":"medizin","score_mean":0.8728670634920634,"score_ci_lo":0.8041666666666666,"score_ci_hi":0.9369295634920635,"top100":87,"n_cells":20,"closed_accuracy":null,"open_rubric_score":0.8728670634920634,"hallu_rate_mean":1.4,"hallu_prevalence":1,"hallu_top100":65,"inter_rater_agreement":0.8018849206349206,"n_judges":4,"score_mean_ex_family":0.8728670634920634,"self_judge_delta":0,"n_cells_ex_family":20,"false_claim_types":{"az":0,"paragraph":0,"study":0,"number":26,"other":60},"tokens_in_total":9096,"tokens_out_total":69730,"model_cost_usd":0.287561},{"model_id":"kimi-k3","domain":"steuerrecht","score_mean":0.6406604308390023,"score_ci_lo":0.41319444444444436,"score_ci_hi":0.792233560090703,"top100":64,"n_cells":7,"closed_accuracy":null,"open_rubric_score":0.6406604308390023,"hallu_rate_mean":3.642857142857143,"hallu_prevalence":1,"hallu_top100":8,"inter_rater_agreement":0.7619614512471655,"n_judges":4,"score_mean_ex_family":0.6406604308390023,"self_judge_delta":0,"n_cells_ex_family":7,"false_claim_types":{"az":0,"paragraph":26,"study":0,"number":14,"other":17},"tokens_in_total":3379,"tokens_out_total":62762,"model_cost_usd":0.254258},{"model_id":"kimi-k3","domain":"wirtschaftsrecht","score_mean":0.731025707384403,"score_ci_lo":0.6164855072463769,"score_ci_hi":0.8180081090407177,"top100":73,"n_cells":23,"closed_accuracy":null,"open_rubric_score":0.731025707384403,"hallu_rate_mean":3.25,"hallu_prevalence":1,"hallu_top100":18,"inter_rater_agreement":0.6392684610075914,"n_judges":4,"score_mean_ex_family":0.731025707384403,"self_judge_delta":0,"n_cells_ex_family":23,"false_claim_types":{"az":1,"paragraph":167,"study":0,"number":18,"other":49},"tokens_in_total":12000,"tokens_out_total":190123,"model_cost_usd":0.771892},{"model_id":"mistral-large-2","domain":"jura","score_mean":0.44948696145124717,"score_ci_lo":0.3661536281179138,"score_ci_hi":0.5298214285714286,"top100":45,"n_cells":35,"closed_accuracy":null,"open_rubric_score":0.44948696145124717,"hallu_rate_mean":2.1357142857142857,"hallu_prevalence":0.8285714285714286,"hallu_top100":46,"inter_rater_agreement":0.6809637188208617,"n_judges":4,"score_mean_ex_family":0.40445578231292517,"self_judge_delta":0.045031179138322,"n_cells_ex_family":35,"false_claim_types":{"az":0,"paragraph":67,"study":0,"number":9,"other":43},"tokens_in_total":13332,"tokens_out_total":10417,"model_cost_usd":0.089166},{"model_id":"mistral-large-2","domain":"medizin","score_mean":0.6815228174603175,"score_ci_lo":0.6315972222222223,"score_ci_hi":0.7353422619047618,"top100":68,"n_cells":20,"closed_accuracy":null,"open_rubric_score":0.6815228174603175,"hallu_rate_mean":1.2875,"hallu_prevalence":0.95,"hallu_top100":67,"inter_rater_agreement":0.6158730158730159,"n_judges":4,"score_mean_ex_family":0.6431547619047618,"self_judge_delta":0.03836805555555567,"n_cells_ex_family":20,"false_claim_types":{"az":0,"paragraph":0,"study":0,"number":25,"other":56},"tokens_in_total":6669,"tokens_out_total":10714,"model_cost_usd":0.077622},{"model_id":"mistral-large-2","domain":"steuerrecht","score_mean":0.29055555555555557,"score_ci_lo":0.1789814814814815,"score_ci_hi":0.4143055555555556,"top100":29,"n_cells":15,"closed_accuracy":null,"open_rubric_score":0.29055555555555557,"hallu_rate_mean":3.3833333333333333,"hallu_prevalence":1,"hallu_top100":15,"inter_rater_agreement":0.5625661375661376,"n_judges":4,"score_mean_ex_family":0.2607407407407407,"self_judge_delta":0.02981481481481485,"n_cells_ex_family":15,"false_claim_types":{"az":0,"paragraph":31,"study":0,"number":31,"other":38},"tokens_in_total":5572,"tokens_out_total":7236,"model_cost_usd":0.05456},{"model_id":"mistral-large-2","domain":"wirtschaftsrecht","score_mean":0.5548214285714285,"score_ci_lo":0.46717261904761903,"score_ci_hi":0.6351785714285715,"top100":55,"n_cells":30,"closed_accuracy":null,"open_rubric_score":0.5548214285714285,"hallu_rate_mean":2.0833333333333335,"hallu_prevalence":1,"hallu_top100":47,"inter_rater_agreement":0.5574603174603175,"n_judges":4,"score_mean_ex_family":0.5100749559082892,"self_judge_delta":0.04474647266313925,"n_cells_ex_family":30,"false_claim_types":{"az":0,"paragraph":92,"study":0,"number":26,"other":57},"tokens_in_total":11870,"tokens_out_total":18300,"model_cost_usd":0.13354}],"by_model_overall":[{"model_id":"claude-fable-5-1","score_mean_across_domains":0.81587200356333,"score_ci_lo":0.7756488905085844,"score_ci_hi":0.8510750728862974,"top100":82,"n_cells":98,"hallu_rate_mean_across_domains":2.260204081632653,"hallu_top100_across_domains":43,"score_mean_ex_family":0.7947238419177195,"self_judge_delta":0.021148161645610508,"n_cells_ex_family":98,"false_claim_types":{"az":15,"paragraph":441,"study":0,"number":54,"other":169}},{"model_id":"claude-opus-4-7","score_mean_across_domains":0.7991071428571429,"score_ci_lo":0.7661369047619047,"score_ci_hi":0.8269593253968255,"top100":80,"n_cells":100,"hallu_rate_mean_across_domains":1.78,"hallu_top100_across_domains":55,"score_mean_ex_family":0.7817480158730159,"self_judge_delta":0.017359126984127005,"n_cells_ex_family":100,"false_claim_types":{"az":17,"paragraph":337,"study":0,"number":65,"other":132}},{"model_id":"claude-opus-4-8","score_mean_across_domains":0.7475525793650794,"score_ci_lo":0.7176696428571429,"score_ci_hi":0.7768353174603174,"top100":75,"n_cells":100,"hallu_rate_mean_across_domains":1.2025,"hallu_top100_across_domains":69,"score_mean_ex_family":0.7196091269841269,"self_judge_delta":0.027943452380952416,"n_cells_ex_family":100,"false_claim_types":{"az":3,"paragraph":210,"study":0,"number":36,"other":118}},{"model_id":"claude-opus-5","score_mean_across_domains":0.8602916666666667,"score_ci_lo":0.8247738095238095,"score_ci_hi":0.8926121031746032,"top100":86,"n_cells":100,"hallu_rate_mean_across_domains":2.3275,"hallu_top100_across_domains":41,"score_mean_ex_family":0.8289047619047619,"self_judge_delta":0.031386904761904755,"n_cells_ex_family":100,"false_claim_types":{"az":17,"paragraph":444,"study":0,"number":74,"other":204}},{"model_id":"deepseek-v4-pro","score_mean_across_domains":0.5980675008098477,"score_ci_lo":0.557597384191772,"score_ci_hi":0.6382693553611921,"top100":60,"n_cells":98,"hallu_rate_mean_across_domains":1.2423469387755102,"hallu_top100_across_domains":68,"score_mean_ex_family":0.5980675008098477,"self_judge_delta":0,"n_cells_ex_family":98,"false_claim_types":{"az":1,"paragraph":130,"study":1,"number":57,"other":85}},{"model_id":"gemini-3-1-pro","score_mean_across_domains":0.5609047619047619,"score_ci_lo":0.5181061507936507,"score_ci_hi":0.601015873015873,"top100":56,"n_cells":100,"hallu_rate_mean_across_domains":0.9625,"hallu_top100_across_domains":75,"score_mean_ex_family":0.5609047619047619,"self_judge_delta":0,"n_cells_ex_family":100,"false_claim_types":{"az":0,"paragraph":53,"study":0,"number":20,"other":126}},{"model_id":"glm-5.3","score_mean_across_domains":0.7867046957671957,"score_ci_lo":0.7395717592592593,"score_ci_hi":0.8274570105820106,"top100":79,"n_cells":60,"hallu_rate_mean_across_domains":1.9125,"hallu_top100_across_domains":52,"score_mean_ex_family":0.7867046957671957,"self_judge_delta":0,"n_cells_ex_family":60,"false_claim_types":{"az":5,"paragraph":168,"study":1,"number":39,"other":114}},{"model_id":"gpt-5","score_mean_across_domains":0.770435515873016,"score_ci_lo":0.7292668650793651,"score_ci_hi":0.8093412698412698,"top100":77,"n_cells":100,"hallu_rate_mean_across_domains":1.2825,"hallu_top100_across_domains":67,"score_mean_ex_family":0.795728835978836,"self_judge_delta":-0.025293320105820083,"n_cells_ex_family":100,"false_claim_types":{"az":0,"paragraph":159,"study":0,"number":67,"other":142}},{"model_id":"gpt-5.6-sol","score_mean_across_domains":0.7601259920634921,"score_ci_lo":0.726577380952381,"score_ci_hi":0.795311507936508,"top100":76,"n_cells":100,"hallu_rate_mean_across_domains":1.4675,"hallu_top100_across_domains":63,"score_mean_ex_family":0.7978862433862433,"self_judge_delta":-0.03776025132275118,"n_cells_ex_family":100,"false_claim_types":{"az":0,"paragraph":188,"study":1,"number":65,"other":135}},{"model_id":"gpt-6-astra","score_mean_across_domains":0.7494047619047619,"score_ci_lo":0.7133670634920635,"score_ci_hi":0.7829375000000001,"top100":75,"n_cells":100,"hallu_rate_mean_across_domains":1.4175,"hallu_top100_across_domains":64,"score_mean_ex_family":0.787702380952381,"self_judge_delta":-0.03829761904761908,"n_cells_ex_family":100,"false_claim_types":{"az":0,"paragraph":180,"study":0,"number":64,"other":122}},{"model_id":"grok-4.5","score_mean_across_domains":0.6643819444444445,"score_ci_lo":0.6315505952380953,"score_ci_hi":0.6946914682539682,"top100":66,"n_cells":100,"hallu_rate_mean_across_domains":0.6575,"hallu_top100_across_domains":83,"score_mean_ex_family":0.6643819444444445,"self_judge_delta":0,"n_cells_ex_family":100,"false_claim_types":{"az":0,"paragraph":56,"study":0,"number":14,"other":82}},{"model_id":"grok-4.6","score_mean_across_domains":0.6278164682539683,"score_ci_lo":0.5878392857142857,"score_ci_hi":0.6675882936507936,"top100":63,"n_cells":100,"hallu_rate_mean_across_domains":1.0375,"hallu_top100_across_domains":74,"score_mean_ex_family":0.6278164682539683,"self_judge_delta":0,"n_cells_ex_family":100,"false_claim_types":{"az":0,"paragraph":104,"study":0,"number":22,"other":112}},{"model_id":"kimi-k3","score_mean_across_domains":0.7763853031172309,"score_ci_lo":0.72220787913559,"score_ci_hi":0.8227720405431249,"top100":78,"n_cells":83,"hallu_rate_mean_across_domains":2.2951807228915664,"hallu_top100_across_domains":42,"score_mean_ex_family":0.7763853031172309,"self_judge_delta":0,"n_cells_ex_family":83,"false_claim_types":{"az":4,"paragraph":311,"study":0,"number":62,"other":159}},{"model_id":"mistral-large-2","score_mean_across_domains":0.5036547619047619,"score_ci_lo":0.4513650793650794,"score_ci_hi":0.5525863095238095,"top100":50,"n_cells":100,"hallu_rate_mean_across_domains":2.1375,"hallu_top100_across_domains":46,"score_mean_ex_family":0.46232407407407405,"self_judge_delta":0.04133068783068783,"n_cells_ex_family":100,"false_claim_types":{"az":0,"paragraph":190,"study":0,"number":91,"other":194}}],"oracle_by_domain":[{"domain":"jura","n_items":35,"oracle_mean":0.9605,"oracle_top100":96,"best_single_model_id":"claude-opus-5","best_single_model_mean":0.8989,"best_single_model_top100":90},{"domain":"medizin","n_items":20,"oracle_mean":0.9649,"oracle_top100":96,"best_single_model_id":"claude-fable-5-1","best_single_model_mean":0.9011,"best_single_model_top100":90},{"domain":"steuerrecht","n_items":15,"oracle_mean":0.9003,"oracle_top100":90,"best_single_model_id":"claude-opus-5","best_single_model_mean":0.8186,"best_single_model_top100":82},{"domain":"wirtschaftsrecht","n_items":30,"oracle_mean":0.9039,"oracle_top100":90,"best_single_model_id":"gpt-5","best_single_model_mean":0.8141,"best_single_model_top100":81}],"cost":{"model_calls_usd":82.6393,"judge_calls_usd":150.4354,"total_usd":233.0747,"by_judge":[{"judge_id":"claude-opus-4-7","tokens_in_total":4178068,"tokens_out_total":851910,"cost_usd":42.18809},{"judge_id":"claude-opus-4-8","tokens_in_total":4171373,"tokens_out_total":1029041,"cost_usd":46.58289},{"judge_id":"gpt-5","tokens_in_total":2148826,"tokens_out_total":2712322,"cost_usd":51.42896},{"judge_id":"mistral-large-2","tokens_in_total":2307394,"tokens_out_total":936785,"cost_usd":10.235498}],"extrapolation":{"scale_factor":12,"extrapolated_total_usd":2796.9,"target_full_run_cells":1200,"observed_cells":100,"note":"linear extrapolation; assumes uniform prompt sizes + judge mix"},"missing_pricing_for":[]}},"raw_jsonl_url":"https://licenses.ai-roundtable.de/r2/runs/run-20260907T181839-5742df/raw.jsonl","progress_jsonl_url":"https://licenses.ai-roundtable.de/r2/runs/run-20260907T181839-5742df/progress.jsonl","cost_total_usd":233.0747}