[{"run_label":"2026-09","arm_key":"anthropic-alias","block":"closed_form_reasoning","calls":600,"cost_usd":0.73593,"input_tokens":52890,"output_tokens":136608},{"run_label":"2026-09","arm_key":"anthropic-alias","block":"instruction_following","calls":300,"cost_usd":0.2553500000000002,"input_tokens":14305,"output_tokens":48209},{"run_label":"2026-09","arm_key":"anthropic-alias","block":"long_context_recall","calls":100,"cost_usd":0.8173649999999999,"input_tokens":806315,"output_tokens":2210},{"run_label":"2026-09","arm_key":"anthropic-alias","block":"multiple_choice","calls":500,"cost_usd":0.5705099999999989,"input_tokens":48710,"output_tokens":104360},{"run_label":"2026-09","arm_key":"anthropic-alias","block":"paraphrase_robustness","calls":200,"cost_usd":0.16369999999999998,"input_tokens":20285,"output_tokens":28683},{"run_label":"2026-09","arm_key":"anthropic-alias","block":"refusal_calibration","calls":200,"cost_usd":0.15000500000000008,"input_tokens":6000,"output_tokens":28801},{"run_label":"2026-09","arm_key":"anthropic-alias","block":"structured_extraction","calls":200,"cost_usd":0.09030000000000002,"input_tokens":37635,"output_tokens":10533},{"run_label":"2026-09","arm_key":"anthropic-snapshot","block":"closed_form_reasoning","calls":600,"cost_usd":0.7356900000000004,"input_tokens":52890,"output_tokens":136560},{"run_label":"2026-09","arm_key":"anthropic-snapshot","block":"instruction_following","calls":300,"cost_usd":0.25617500000000004,"input_tokens":14305,"output_tokens":48374},{"run_label":"2026-09","arm_key":"anthropic-snapshot","block":"long_context_recall","calls":100,"cost_usd":0.8173649999999999,"input_tokens":806315,"output_tokens":2210},{"run_label":"2026-09","arm_key":"anthropic-snapshot","block":"multiple_choice","calls":500,"cost_usd":0.5717599999999994,"input_tokens":48710,"output_tokens":104610},{"run_label":"2026-09","arm_key":"anthropic-snapshot","block":"paraphrase_robustness","calls":200,"cost_usd":0.16411499999999993,"input_tokens":20285,"output_tokens":28766},{"run_label":"2026-09","arm_key":"anthropic-snapshot","block":"refusal_calibration","calls":200,"cost_usd":0.15103000000000008,"input_tokens":6000,"output_tokens":29006},{"run_label":"2026-09","arm_key":"anthropic-snapshot","block":"structured_extraction","calls":200,"cost_usd":0.09030000000000005,"input_tokens":37635,"output_tokens":10533},{"run_label":"2026-09","arm_key":"anthropic-sonnet-snapshot","block":"closed_form_reasoning","calls":600,"cost_usd":1.0760599999999996,"input_tokens":65815,"output_tokens":94443},{"run_label":"2026-09","arm_key":"anthropic-sonnet-snapshot","block":"instruction_following","calls":300,"cost_usd":1.0762899999999997,"input_tokens":20035,"output_tokens":103622},{"run_label":"2026-09","arm_key":"anthropic-sonnet-snapshot","block":"long_context_recall","calls":100,"cost_usd":2.1399399999999997,"input_tokens":1066430,"output_tokens":708},{"run_label":"2026-09","arm_key":"anthropic-sonnet-snapshot","block":"multiple_choice","calls":500,"cost_usd":0.28309999999999985,"input_tokens":64815,"output_tokens":15347},{"run_label":"2026-09","arm_key":"anthropic-sonnet-snapshot","block":"paraphrase_robustness","calls":200,"cost_usd":0.35527000000000003,"input_tokens":26035,"output_tokens":30320},{"run_label":"2026-09","arm_key":"anthropic-sonnet-snapshot","block":"refusal_calibration","calls":200,"cost_usd":0.6927400000000002,"input_tokens":8415,"output_tokens":67591},{"run_label":"2026-09","arm_key":"anthropic-sonnet-snapshot","block":"structured_extraction","calls":200,"cost_usd":0.20920999999999998,"input_tokens":49155,"output_tokens":11090},{"run_label":"2026-09","arm_key":"google-alias","block":"closed_form_reasoning","calls":600,"cost_usd":0.6101474999999996,"input_tokens":45820,"output_tokens":153542},{"run_label":"2026-09","arm_key":"google-alias","block":"instruction_following","calls":300,"cost_usd":0.6552187499999998,"input_tokens":12045,"output_tokens":172316},{"run_label":"2026-09","arm_key":"google-alias","block":"long_context_recall","calls":100,"cost_usd":0.48027180000000025,"input_tokens":617056,"output_tokens":2110},{"run_label":"2026-09","arm_key":"google-alias","block":"multiple_choice","calls":500,"cost_usd":0.24803250000000018,"input_tokens":42925,"output_tokens":57557},{"run_label":"2026-09","arm_key":"google-alias","block":"paraphrase_robustness","calls":200,"cost_usd":0.1712287500000002,"input_tokens":17995,"output_tokens":42062},{"run_label":"2026-09","arm_key":"google-alias","block":"refusal_calibration","calls":200,"cost_usd":0.20461875,"input_tokens":4550,"output_tokens":53655},{"run_label":"2026-09","arm_key":"google-alias","block":"structured_extraction","calls":200,"cost_usd":0.06067499999999997,"input_tokens":35855,"output_tokens":9009},{"run_label":"2026-09","arm_key":"google-snapshot","block":"closed_form_reasoning","calls":600,"cost_usd":0.6050999999999997,"input_tokens":45820,"output_tokens":152196},{"run_label":"2026-09","arm_key":"google-snapshot","block":"instruction_following","calls":300,"cost_usd":0.6118815000000001,"input_tokens":12002,"output_tokens":160768},{"run_label":"2026-09","arm_key":"google-snapshot","block":"long_context_recall","calls":100,"cost_usd":0.4893962250000001,"input_tokens":633007,"output_tokens":1672},{"run_label":"2026-09","arm_key":"google-snapshot","block":"multiple_choice","calls":500,"cost_usd":0.23992875000000005,"input_tokens":43050,"output_tokens":55371},{"run_label":"2026-09","arm_key":"google-snapshot","block":"paraphrase_robustness","calls":200,"cost_usd":0.1695937500000001,"input_tokens":17995,"output_tokens":41626},{"run_label":"2026-09","arm_key":"google-snapshot","block":"refusal_calibration","calls":200,"cost_usd":0.19796624999999987,"input_tokens":4550,"output_tokens":51881},{"run_label":"2026-09","arm_key":"google-snapshot","block":"structured_extraction","calls":200,"cost_usd":0.06129374999999997,"input_tokens":36030,"output_tokens":9139},{"run_label":"2026-09","arm_key":"openai-alias","block":"closed_form_reasoning","calls":600,"cost_usd":0.2843759999999999,"input_tokens":50260,"output_tokens":54818},{"run_label":"2026-09","arm_key":"openai-alias","block":"instruction_following","calls":300,"cost_usd":0.17535749999999992,"input_tokens":14510,"output_tokens":36550},{"run_label":"2026-09","arm_key":"openai-alias","block":"long_context_recall","calls":100,"cost_usd":0.2337884999999999,"input_tokens":260920,"output_tokens":573},{"run_label":"2026-09","arm_key":"openai-alias","block":"multiple_choice","calls":500,"cost_usd":0.04428000000000003,"input_tokens":46500,"output_tokens":2090},{"run_label":"2026-09","arm_key":"openai-alias","block":"paraphrase_robustness","calls":200,"cost_usd":0.09156899999999998,"input_tokens":19360,"output_tokens":17122},{"run_label":"2026-09","arm_key":"openai-alias","block":"refusal_calibration","calls":200,"cost_usd":0.13937325000000003,"input_tokens":6305,"output_tokens":29921},{"run_label":"2026-09","arm_key":"openai-alias","block":"structured_extraction","calls":200,"cost_usd":0.05326874999999999,"input_tokens":35925,"output_tokens":5850},{"run_label":"2026-09","arm_key":"openai-snapshot","block":"closed_form_reasoning","calls":600,"cost_usd":0.2860860000000001,"input_tokens":50260,"output_tokens":55198},{"run_label":"2026-09","arm_key":"openai-snapshot","block":"instruction_following","calls":300,"cost_usd":0.18051899999999993,"input_tokens":14510,"output_tokens":37697},{"run_label":"2026-09","arm_key":"openai-snapshot","block":"long_context_recall","calls":100,"cost_usd":0.24311519999999984,"input_tokens":274744,"output_tokens":572},{"run_label":"2026-09","arm_key":"openai-snapshot","block":"multiple_choice","calls":500,"cost_usd":0.044145000000000024,"input_tokens":46500,"output_tokens":2060},{"run_label":"2026-09","arm_key":"openai-snapshot","block":"paraphrase_robustness","calls":200,"cost_usd":0.09115949999999996,"input_tokens":19360,"output_tokens":17031},{"run_label":"2026-09","arm_key":"openai-snapshot","block":"refusal_calibration","calls":200,"cost_usd":0.13851824999999995,"input_tokens":6305,"output_tokens":29731},{"run_label":"2026-09","arm_key":"openai-snapshot","block":"structured_extraction","calls":200,"cost_usd":0.053291249999999984,"input_tokens":35925,"output_tokens":5855},{"run_label":"2026-09","arm_key":"openweights-control","block":"closed_form_reasoning","calls":600,"cost_usd":0.20841807999999995,"input_tokens":65717,"output_tokens":134685},{"run_label":"2026-09","arm_key":"openweights-control","block":"instruction_following","calls":300,"cost_usd":0.06456320000000008,"input_tokens":22100,"output_tokens":39980},{"run_label":"2026-09","arm_key":"openweights-control","block":"long_context_recall","calls":100,"cost_usd":0.7743215999999999,"input_tokens":744165,"output_tokens":375},{"run_label":"2026-09","arm_key":"openweights-control","block":"multiple_choice","calls":500,"cost_usd":0.06995663999999999,"input_tokens":59325,"output_tokens":7941},{"run_label":"2026-09","arm_key":"openweights-control","block":"paraphrase_robustness","calls":200,"cost_usd":0.05183984000000002,"input_tokens":24605,"output_tokens":25241},{"run_label":"2026-09","arm_key":"openweights-control","block":"refusal_calibration","calls":200,"cost_usd":0.043125680000000034,"input_tokens":11360,"output_tokens":30107},{"run_label":"2026-09","arm_key":"openweights-control","block":"structured_extraction","calls":200,"cost_usd":0.051940720000000086,"input_tokens":41080,"output_tokens":8863},{"run_label":"2026-09-run2","arm_key":"anthropic-alias","block":"closed_form_reasoning","calls":600,"cost_usd":0.7312249999999995,"input_tokens":52890,"output_tokens":135667},{"run_label":"2026-09-run2","arm_key":"anthropic-alias","block":"instruction_following","calls":300,"cost_usd":0.2559000000000002,"input_tokens":14305,"output_tokens":48319},{"run_label":"2026-09-run2","arm_key":"anthropic-alias","block":"long_context_recall","calls":100,"cost_usd":0.8173649999999999,"input_tokens":806315,"output_tokens":2210},{"run_label":"2026-09-run2","arm_key":"anthropic-alias","block":"multiple_choice","calls":500,"cost_usd":0.5833149999999993,"input_tokens":48710,"output_tokens":106921},{"run_label":"2026-09-run2","arm_key":"anthropic-alias","block":"paraphrase_robustness","calls":200,"cost_usd":0.16340000000000007,"input_tokens":20285,"output_tokens":28623},{"run_label":"2026-09-run2","arm_key":"anthropic-alias","block":"refusal_calibration","calls":200,"cost_usd":0.14904000000000003,"input_tokens":6000,"output_tokens":28608},{"run_label":"2026-09-run2","arm_key":"anthropic-alias","block":"structured_extraction","calls":200,"cost_usd":0.0903,"input_tokens":37635,"output_tokens":10533},{"run_label":"2026-09-run2","arm_key":"anthropic-snapshot","block":"closed_form_reasoning","calls":600,"cost_usd":0.7382250000000004,"input_tokens":52890,"output_tokens":137067},{"run_label":"2026-09-run2","arm_key":"anthropic-snapshot","block":"instruction_following","calls":300,"cost_usd":0.25602000000000025,"input_tokens":14305,"output_tokens":48343},{"run_label":"2026-09-run2","arm_key":"anthropic-snapshot","block":"long_context_recall","calls":100,"cost_usd":0.8173649999999999,"input_tokens":806315,"output_tokens":2210},{"run_label":"2026-09-run2","arm_key":"anthropic-snapshot","block":"multiple_choice","calls":500,"cost_usd":0.5798249999999996,"input_tokens":48710,"output_tokens":106223},{"run_label":"2026-09-run2","arm_key":"anthropic-snapshot","block":"paraphrase_robustness","calls":200,"cost_usd":0.16407499999999986,"input_tokens":20285,"output_tokens":28758},{"run_label":"2026-09-run2","arm_key":"anthropic-snapshot","block":"refusal_calibration","calls":200,"cost_usd":0.14982000000000006,"input_tokens":6000,"output_tokens":28764},{"run_label":"2026-09-run2","arm_key":"anthropic-snapshot","block":"structured_extraction","calls":200,"cost_usd":0.09029500000000004,"input_tokens":37635,"output_tokens":10532},{"run_label":"2026-09-run2","arm_key":"anthropic-sonnet-snapshot","block":"closed_form_reasoning","calls":600,"cost_usd":1.0861299999999998,"input_tokens":65815,"output_tokens":95450},{"run_label":"2026-09-run2","arm_key":"anthropic-sonnet-snapshot","block":"instruction_following","calls":300,"cost_usd":1.131490000000001,"input_tokens":20035,"output_tokens":109142},{"run_label":"2026-09-run2","arm_key":"anthropic-sonnet-snapshot","block":"long_context_recall","calls":100,"cost_usd":2.1395,"input_tokens":1066430,"output_tokens":664},{"run_label":"2026-09-run2","arm_key":"anthropic-sonnet-snapshot","block":"multiple_choice","calls":500,"cost_usd":0.28621999999999986,"input_tokens":64815,"output_tokens":15659},{"run_label":"2026-09-run2","arm_key":"anthropic-sonnet-snapshot","block":"paraphrase_robustness","calls":200,"cost_usd":0.3537799999999999,"input_tokens":26035,"output_tokens":30171},{"run_label":"2026-09-run2","arm_key":"anthropic-sonnet-snapshot","block":"refusal_calibration","calls":200,"cost_usd":0.7006799999999999,"input_tokens":8415,"output_tokens":68385},{"run_label":"2026-09-run2","arm_key":"anthropic-sonnet-snapshot","block":"structured_extraction","calls":200,"cost_usd":0.20843,"input_tokens":49155,"output_tokens":11012},{"run_label":"2026-09-run2","arm_key":"google-alias","block":"closed_form_reasoning","calls":600,"cost_usd":0.6036149999999997,"input_tokens":45820,"output_tokens":151800},{"run_label":"2026-09-run2","arm_key":"google-alias","block":"instruction_following","calls":300,"cost_usd":0.6445004999999995,"input_tokens":12004,"output_tokens":169466},{"run_label":"2026-09-run2","arm_key":"google-alias","block":"long_context_recall","calls":100,"cost_usd":0.49446975,"input_tokens":642740,"output_tokens":1273},{"run_label":"2026-09-run2","arm_key":"google-alias","block":"multiple_choice","calls":500,"cost_usd":0.2295450000000001,"input_tokens":43050,"output_tokens":52602},{"run_label":"2026-09-run2","arm_key":"google-alias","block":"paraphrase_robustness","calls":200,"cost_usd":0.16991999999999996,"input_tokens":17995,"output_tokens":41713},{"run_label":"2026-09-run2","arm_key":"google-alias","block":"refusal_calibration","calls":200,"cost_usd":0.20262525000000003,"input_tokens":4522,"output_tokens":53129},{"run_label":"2026-09-run2","arm_key":"google-alias","block":"structured_extraction","calls":200,"cost_usd":0.06128250000000002,"input_tokens":36030,"output_tokens":9136},{"run_label":"2026-09-run2","arm_key":"google-snapshot","block":"closed_form_reasoning","calls":600,"cost_usd":0.6079349999999988,"input_tokens":45820,"output_tokens":152952},{"run_label":"2026-09-run2","arm_key":"google-snapshot","block":"instruction_following","calls":300,"cost_usd":0.6554362499999992,"input_tokens":12045,"output_tokens":172374},{"run_label":"2026-09-run2","arm_key":"google-snapshot","block":"long_context_recall","calls":100,"cost_usd":0.4927752750000004,"input_tokens":638663,"output_tokens":1555},{"run_label":"2026-09-run2","arm_key":"google-snapshot","block":"multiple_choice","calls":500,"cost_usd":0.23727375000000017,"input_tokens":43050,"output_tokens":54663},{"run_label":"2026-09-run2","arm_key":"google-snapshot","block":"paraphrase_robustness","calls":200,"cost_usd":0.17210625000000002,"input_tokens":17995,"output_tokens":42296},{"run_label":"2026-09-run2","arm_key":"google-snapshot","block":"refusal_calibration","calls":200,"cost_usd":0.19480125000000006,"input_tokens":4550,"output_tokens":51037},{"run_label":"2026-09-run2","arm_key":"google-snapshot","block":"structured_extraction","calls":200,"cost_usd":0.06141750000000002,"input_tokens":36030,"output_tokens":9172},{"run_label":"2026-09-run2","arm_key":"openai-alias","block":"closed_form_reasoning","calls":600,"cost_usd":0.28570799999999985,"input_tokens":50260,"output_tokens":55114},{"run_label":"2026-09-run2","arm_key":"openai-alias","block":"instruction_following","calls":300,"cost_usd":0.17222999999999994,"input_tokens":14510,"output_tokens":35855},{"run_label":"2026-09-run2","arm_key":"openai-alias","block":"long_context_recall","calls":100,"cost_usd":0.21511709999999992,"input_tokens":233272,"output_tokens":571},{"run_label":"2026-09-run2","arm_key":"openai-alias","block":"multiple_choice","calls":500,"cost_usd":0.04425300000000003,"input_tokens":46500,"output_tokens":2084},{"run_label":"2026-09-run2","arm_key":"openai-alias","block":"paraphrase_robustness","calls":200,"cost_usd":0.09041700000000003,"input_tokens":19360,"output_tokens":16866},{"run_label":"2026-09-run2","arm_key":"openai-alias","block":"refusal_calibration","calls":200,"cost_usd":0.13645274999999984,"input_tokens":6305,"output_tokens":29272},{"run_label":"2026-09-run2","arm_key":"openai-alias","block":"structured_extraction","calls":200,"cost_usd":0.05325974999999999,"input_tokens":35925,"output_tokens":5848},{"run_label":"2026-09-run2","arm_key":"openai-snapshot","block":"closed_form_reasoning","calls":600,"cost_usd":0.2855369999999999,"input_tokens":50260,"output_tokens":55076},{"run_label":"2026-09-run2","arm_key":"openai-snapshot","block":"instruction_following","calls":300,"cost_usd":0.17559149999999996,"input_tokens":14510,"output_tokens":36602},{"run_label":"2026-09-run2","arm_key":"openai-snapshot","block":"long_context_recall","calls":100,"cost_usd":0.20544479999999987,"input_tokens":218936,"output_tokens":572},{"run_label":"2026-09-run2","arm_key":"openai-snapshot","block":"multiple_choice","calls":500,"cost_usd":0.044334000000000026,"input_tokens":46500,"output_tokens":2102},{"run_label":"2026-09-run2","arm_key":"openai-snapshot","block":"paraphrase_robustness","calls":200,"cost_usd":0.09108749999999997,"input_tokens":19360,"output_tokens":17015},{"run_label":"2026-09-run2","arm_key":"openai-snapshot","block":"refusal_calibration","calls":200,"cost_usd":0.13441874999999992,"input_tokens":6305,"output_tokens":28820},{"run_label":"2026-09-run2","arm_key":"openai-snapshot","block":"structured_extraction","calls":200,"cost_usd":0.05322374999999998,"input_tokens":35925,"output_tokens":5840},{"run_label":"2026-09-run2","arm_key":"openweights-control","block":"closed_form_reasoning","calls":600,"cost_usd":0.21106176000000007,"input_tokens":65837,"output_tokens":137107},{"run_label":"2026-09-run2","arm_key":"openweights-control","block":"instruction_following","calls":300,"cost_usd":0.06256328000000003,"input_tokens":21815,"output_tokens":38342},{"run_label":"2026-09-run2","arm_key":"openweights-control","block":"long_context_recall","calls":100,"cost_usd":0.7743215999999999,"input_tokens":744165,"output_tokens":375},{"run_label":"2026-09-run2","arm_key":"openweights-control","block":"multiple_choice","calls":500,"cost_usd":0.07119839999999997,"input_tokens":59325,"output_tokens":9135},{"run_label":"2026-09-run2","arm_key":"openweights-control","block":"paraphrase_robustness","calls":200,"cost_usd":0.05343624000000001,"input_tokens":24605,"output_tokens":26776},{"run_label":"2026-09-run2","arm_key":"openweights-control","block":"refusal_calibration","calls":200,"cost_usd":0.041177760000000015,"input_tokens":11312,"output_tokens":28282},{"run_label":"2026-09-run2","arm_key":"openweights-control","block":"structured_extraction","calls":200,"cost_usd":0.05190848000000008,"input_tokens":41080,"output_tokens":8832}]