{
  "date": "2026-09-22",
  "findings": 2449,
  "severityDistribution": {
    "medium": 932,
    "high": 330,
    "info": 1002,
    "critical": 55,
    "low": 130
  },
  "corpusSha256": "b94f06b0e793c8ae21879c1133a9d783e3cb5f75e58e5623ba62e41246a3cfd0",
  "runSha256": "c37fdd6d3abd7f73839115a0793caf3449b634c93d0697e1e932b774296fce24",
  "interpretation": "Agreement with Casco published CVSS 3.1 decisions.",
  "models": [
    {
      "id": "gpt-6-astra",
      "name": "GPT-6 Astra",
      "setting": "Low reasoning",
      "all": {
        "n": 2449,
        "mae": 1.9179256839526337,
        "mean_signed_error": 1.1772968558595345,
        "severity_accuracy_pct": 53.368721927317274,
        "info_overstatement_pct": 65.06986027944112,
        "info_denominator": 1002,
        "false_critical_pct": 1.545530492898914,
        "false_critical_denominator": 2394,
        "critical_recall_pct": 60.0,
        "under_score_pct": 20.253164556962027,
        "metric_accuracy_pct": {
          "AV": 94.52837893017558,
          "AC": 80.23683135973867,
          "PR": 93.22172315230706,
          "UI": 76.27603103307473,
          "S": 85.34095549203757,
          "C": 64.23029808084932,
          "I": 72.35606369946917,
          "A": 92.28256431196407
        },
        "mae_by_reference_severity": {
          "info": 3.1788423153692613,
          "low": 1.953076923076923,
          "medium": 0.8188841201716739,
          "high": 1.3806060606060606,
          "critical": 0.7109090909090909
        },
        "predicted_distribution": {
          "info": 378,
          "medium": 1510,
          "high": 314,
          "low": 177,
          "critical": 70
        }
      },
      "maeInterval": {
        "lower": 1.8022616407982261,
        "upper": 2.0282894736842105,
        "applications": 161,
        "draws": 2000,
        "seed": 20260922
      },
      "infoEscalations": 652,
      "falseCritical": 37
    },
    {
      "id": "opus",
      "name": "Claude Opus 4.6",
      "setting": "No extended thinking",
      "all": {
        "n": 2449,
        "mae": 2.08705594120049,
        "mean_signed_error": 1.676194365046958,
        "severity_accuracy_pct": 47.039608003266636,
        "info_overstatement_pct": 83.43313373253493,
        "info_denominator": 1002,
        "false_critical_pct": 2.046783625730994,
        "false_critical_denominator": 2394,
        "critical_recall_pct": 74.54545454545455,
        "under_score_pct": 13.924050632911392,
        "metric_accuracy_pct": {
          "AV": 98.28501429154757,
          "AC": 83.95263372805226,
          "PR": 93.75255206206614,
          "UI": 81.74765210289914,
          "S": 87.83176806859943,
          "C": 64.96529195590037,
          "I": 68.68109432421396,
          "A": 91.99673336055533
        },
        "mae_by_reference_severity": {
          "info": 3.7972055888223553,
          "low": 1.9653846153846153,
          "medium": 0.73068669527897,
          "high": 1.0166666666666666,
          "critical": 0.6254545454545455
        },
        "predicted_distribution": {
          "low": 276,
          "info": 173,
          "high": 372,
          "critical": 90,
          "medium": 1538
        }
      },
      "maeInterval": {
        "lower": 1.9586531226486081,
        "upper": 2.222471416007036,
        "applications": 161,
        "draws": 2000,
        "seed": 20260922
      },
      "infoEscalations": 836,
      "falseCritical": 49
    },
    {
      "id": "gpt-5.6-terra",
      "name": "GPT-5.6 Terra",
      "setting": "Reasoning disabled",
      "all": {
        "n": 2449,
        "mae": 2.174438546345447,
        "mean_signed_error": 1.713679052674561,
        "severity_accuracy_pct": 48.999591670069414,
        "info_overstatement_pct": 70.15968063872255,
        "info_denominator": 1002,
        "false_critical_pct": 4.218880534670008,
        "false_critical_denominator": 2394,
        "critical_recall_pct": 67.27272727272727,
        "under_score_pct": 15.108207431604736,
        "metric_accuracy_pct": {
          "AV": 91.99673336055533,
          "AC": 76.8476929358922,
          "PR": 93.01755818701511,
          "UI": 78.56267864434463,
          "S": 85.54512045732952,
          "C": 62.9644752960392,
          "I": 72.80522662311148,
          "A": 72.35606369946917
        },
        "mae_by_reference_severity": {
          "info": 3.7501996007984033,
          "low": 2.1884615384615382,
          "medium": 0.9563304721030044,
          "high": 1.0730303030303032,
          "critical": 0.6836363636363637
        },
        "predicted_distribution": {
          "info": 320,
          "medium": 1397,
          "high": 468,
          "low": 126,
          "critical": 138
        }
      },
      "maeInterval": {
        "lower": 2.0328414839797637,
        "upper": 2.3129843363561418,
        "applications": 161,
        "draws": 2000,
        "seed": 20260922
      },
      "infoEscalations": 703,
      "falseCritical": 101
    },
    {
      "id": "gpt-5.6-sol",
      "name": "GPT-5.6 Sol",
      "setting": "Reasoning disabled",
      "all": {
        "n": 2449,
        "mae": 2.347080440996325,
        "mean_signed_error": 1.847611269906084,
        "severity_accuracy_pct": 44.99795835034708,
        "info_overstatement_pct": 84.83033932135729,
        "info_denominator": 1002,
        "false_critical_pct": 2.3391812865497075,
        "false_critical_denominator": 2394,
        "critical_recall_pct": 61.81818181818182,
        "under_score_pct": 17.231523070641078,
        "metric_accuracy_pct": {
          "AV": 92.65006124948958,
          "AC": 81.13515720702327,
          "PR": 91.13924050632912,
          "UI": 72.31523070641077,
          "S": 85.50428746427113,
          "C": 58.59534503879134,
          "I": 63.82196815026541,
          "A": 83.99346672111066
        },
        "mae_by_reference_severity": {
          "info": 4.256387225548902,
          "low": 1.9953846153846153,
          "medium": 0.8664163090128756,
          "high": 1.1572727272727272,
          "critical": 0.6236363636363637
        },
        "predicted_distribution": {
          "info": 158,
          "medium": 1562,
          "high": 417,
          "low": 222,
          "critical": 90
        }
      },
      "maeInterval": {
        "lower": 2.211710261569417,
        "upper": 2.483380547686497,
        "applications": 161,
        "draws": 2000,
        "seed": 20260922
      },
      "infoEscalations": 850,
      "falseCritical": 56
    },
    {
      "id": "sonnet",
      "name": "Claude Sonnet 4.6",
      "setting": "No extended thinking",
      "all": {
        "n": 2449,
        "mae": 2.4392813393221724,
        "mean_signed_error": 2.0815026541445487,
        "severity_accuracy_pct": 42.752143732135565,
        "info_overstatement_pct": 94.81037924151697,
        "info_denominator": 1002,
        "false_critical_pct": 3.007518796992481,
        "false_critical_denominator": 2394,
        "critical_recall_pct": 74.54545454545455,
        "under_score_pct": 12.862392813393221,
        "metric_accuracy_pct": {
          "AV": 95.71253572886893,
          "AC": 82.31931400571662,
          "PR": 93.26255614536545,
          "UI": 77.70518579011842,
          "S": 81.99265006124949,
          "C": 55.90036749693753,
          "I": 64.96529195590037,
          "A": 91.54757043691302
        },
        "mae_by_reference_severity": {
          "info": 4.605688622754491,
          "low": 2.299230769230769,
          "medium": 0.7685622317596567,
          "high": 0.9309090909090909,
          "critical": 0.6636363636363638
        },
        "predicted_distribution": {
          "info": 54,
          "medium": 1654,
          "critical": 113,
          "high": 397,
          "low": 231
        }
      },
      "maeInterval": {
        "lower": 2.302374245472837,
        "upper": 2.5719838465421505,
        "applications": 161,
        "draws": 2000,
        "seed": 20260922
      },
      "infoEscalations": 950,
      "falseCritical": 72
    },
    {
      "id": "gpt-5.6-luna",
      "name": "GPT-5.6 Luna",
      "setting": "Reasoning disabled",
      "all": {
        "n": 2449,
        "mae": 2.7569211923233974,
        "mean_signed_error": 2.4991016741527154,
        "severity_accuracy_pct": 40.46549612086566,
        "info_overstatement_pct": 80.73852295409182,
        "info_denominator": 1002,
        "false_critical_pct": 11.487050960735171,
        "false_critical_denominator": 2394,
        "critical_recall_pct": 87.27272727272727,
        "under_score_pct": 9.39158840342997,
        "metric_accuracy_pct": {
          "AV": 97.10085749285423,
          "AC": 74.0710494079216,
          "PR": 92.65006124948958,
          "UI": 78.93017558187015,
          "S": 82.15598203348306,
          "C": 55.69620253164557,
          "I": 62.84197631686403,
          "A": 72.96855859534504
        },
        "mae_by_reference_severity": {
          "info": 4.7604790419161676,
          "low": 2.9846153846153847,
          "medium": 1.284442060085837,
          "high": 1.1327272727272728,
          "critical": 0.4145454545454547
        },
        "predicted_distribution": {
          "medium": 1323,
          "high": 556,
          "critical": 323,
          "info": 196,
          "low": 51
        }
      },
      "maeInterval": {
        "lower": 2.5939858050056035,
        "upper": 2.9252459016393444,
        "applications": 161,
        "draws": 2000,
        "seed": 20260922
      },
      "infoEscalations": 809,
      "falseCritical": 275
    },
    {
      "id": "jev",
      "name": "Jev 1.13.0",
      "setting": "Eight Choice questions",
      "all": {
        "n": 2449,
        "mae": 3.3958758677011023,
        "mean_signed_error": 3.299591670069416,
        "severity_accuracy_pct": 30.78807676602695,
        "info_overstatement_pct": 83.63273453093812,
        "info_denominator": 1002,
        "false_critical_pct": 20.885547201336674,
        "false_critical_denominator": 2394,
        "critical_recall_pct": 90.9090909090909,
        "under_score_pct": 3.144140465496121,
        "metric_accuracy_pct": {
          "AV": 98.48917925683952,
          "AC": 65.04695794201714,
          "PR": 89.54675377705186,
          "UI": 87.79093507554104,
          "S": 83.09514087382605,
          "C": 50.18374846876276,
          "I": 59.82033483054308,
          "A": 90.03674969375255
        },
        "mae_by_reference_severity": {
          "info": 5.662275449101796,
          "low": 3.859230769230769,
          "medium": 1.9627682403433475,
          "high": 0.8721212121212122,
          "critical": 0.4381818181818184
        },
        "predicted_distribution": {
          "medium": 879,
          "info": 180,
          "critical": 550,
          "high": 826,
          "low": 14
        }
      },
      "maeInterval": {
        "lower": 3.217072129748186,
        "upper": 3.55791788856305,
        "applications": 161,
        "draws": 2000,
        "seed": 20260922
      },
      "infoEscalations": 838,
      "falseCritical": 500
    },
    {
      "id": "haiku",
      "name": "Claude Haiku 4.5",
      "setting": "No extended thinking",
      "all": {
        "n": 2449,
        "mae": 3.936014699877501,
        "mean_signed_error": 3.8854634544712128,
        "severity_accuracy_pct": 25.60228664761127,
        "info_overstatement_pct": 98.40319361277444,
        "info_denominator": 1002,
        "false_critical_pct": 19.46532999164578,
        "false_critical_denominator": 2394,
        "critical_recall_pct": 94.54545454545455,
        "under_score_pct": 2.5724785626786444,
        "metric_accuracy_pct": {
          "AV": 98.44834626378113,
          "AC": 69.82441812984892,
          "PR": 91.99673336055533,
          "UI": 78.88934258881176,
          "S": 80.80849326255614,
          "C": 34.95304205798285,
          "I": 54.512045732952224,
          "A": 84.03429971416905
        },
        "mae_by_reference_severity": {
          "info": 6.984431137724551,
          "low": 3.9076923076923076,
          "medium": 1.9580472103004292,
          "high": 0.8654545454545456,
          "critical": 0.40727272727272745
        },
        "predicted_distribution": {
          "high": 1005,
          "medium": 893,
          "critical": 518,
          "info": 17,
          "low": 16
        }
      },
      "maeInterval": {
        "lower": 3.7250900900900903,
        "upper": 4.136701883325678,
        "applications": 161,
        "draws": 2000,
        "seed": 20260922
      },
      "infoEscalations": 986,
      "falseCritical": 466
    }
  ],
  "categories": [
    {
      "name": "Security misconfiguration",
      "n": 879,
      "models": {
        "jev": {
          "mae": 4.534926052332195,
          "signedError": 4.510807736063708,
          "overstated": 695,
          "infoCount": 655,
          "infoEscalated": 537
        },
        "haiku": {
          "mae": 5.710580204778157,
          "signedError": 5.696018202502844,
          "overstated": 818,
          "infoCount": 655,
          "infoEscalated": 643
        },
        "sonnet": {
          "mae": 3.4425483503981797,
          "signedError": 3.3381114903299203,
          "overstated": 681,
          "infoCount": 655,
          "infoEscalated": 620
        },
        "opus": {
          "mae": 2.735836177474403,
          "signedError": 2.574744027303754,
          "overstated": 571,
          "infoCount": 655,
          "infoEscalated": 524
        },
        "gpt-6-astra": {
          "mae": 2.2814562002275314,
          "signedError": 2.054835039817975,
          "overstated": 435,
          "infoCount": 655,
          "infoEscalated": 397
        },
        "gpt-5.6-sol": {
          "mae": 3.181797497155859,
          "signedError": 3.0448236632536974,
          "overstated": 604,
          "infoCount": 655,
          "infoEscalated": 551
        },
        "gpt-5.6-terra": {
          "mae": 2.6629124004550624,
          "signedError": 2.4560864618885097,
          "overstated": 475,
          "infoCount": 655,
          "infoEscalated": 423
        },
        "gpt-5.6-luna": {
          "mae": 3.6006825938566553,
          "signedError": 3.512172923777019,
          "overstated": 589,
          "infoCount": 655,
          "infoEscalated": 514
        }
      }
    },
    {
      "name": "Access control / authorization",
      "n": 656,
      "models": {
        "jev": {
          "mae": 1.958689024390244,
          "signedError": 1.8696646341463414,
          "overstated": 459,
          "infoCount": 52,
          "infoEscalated": 47
        },
        "haiku": {
          "mae": 1.9332317073170733,
          "signedError": 1.8390243902439025,
          "overstated": 450,
          "infoCount": 52,
          "infoEscalated": 51
        },
        "sonnet": {
          "mae": 1.2452743902439023,
          "signedError": 0.5364329268292682,
          "overstated": 248,
          "infoCount": 52,
          "infoEscalated": 50
        },
        "opus": {
          "mae": 1.1704268292682927,
          "signedError": 0.49878048780487805,
          "overstated": 240,
          "infoCount": 52,
          "infoEscalated": 50
        },
        "gpt-6-astra": {
          "mae": 1.275762195121951,
          "signedError": -0.09131097560975612,
          "overstated": 170,
          "infoCount": 52,
          "infoEscalated": 42
        },
        "gpt-5.6-sol": {
          "mae": 1.2777439024390245,
          "signedError": 0.3594512195121951,
          "overstated": 232,
          "infoCount": 52,
          "infoEscalated": 50
        },
        "gpt-5.6-terra": {
          "mae": 1.2789634146341464,
          "signedError": 0.49573170731707317,
          "overstated": 271,
          "infoCount": 52,
          "infoEscalated": 47
        },
        "gpt-5.6-luna": {
          "mae": 1.4695121951219512,
          "signedError": 0.9384146341463415,
          "overstated": 347,
          "infoCount": 52,
          "infoEscalated": 49
        }
      }
    },
    {
      "name": "Other / unclassified",
      "n": 300,
      "models": {
        "jev": {
          "mae": 3.8706666666666667,
          "signedError": 3.76,
          "overstated": 220,
          "infoCount": 150,
          "infoEscalated": 121
        },
        "haiku": {
          "mae": 4.650666666666667,
          "signedError": 4.614666666666667,
          "overstated": 262,
          "infoCount": 150,
          "infoEscalated": 147
        },
        "sonnet": {
          "mae": 3.0673333333333335,
          "signedError": 2.742,
          "overstated": 186,
          "infoCount": 150,
          "infoEscalated": 137
        },
        "opus": {
          "mae": 2.592,
          "signedError": 2.1613333333333333,
          "overstated": 171,
          "infoCount": 150,
          "infoEscalated": 127
        },
        "gpt-6-astra": {
          "mae": 2.5816666666666666,
          "signedError": 1.8443333333333332,
          "overstated": 143,
          "infoCount": 150,
          "infoEscalated": 105
        },
        "gpt-5.6-sol": {
          "mae": 2.8926666666666665,
          "signedError": 2.4193333333333333,
          "overstated": 173,
          "infoCount": 150,
          "infoEscalated": 120
        },
        "gpt-5.6-terra": {
          "mae": 2.966333333333333,
          "signedError": 2.4703333333333335,
          "overstated": 168,
          "infoCount": 150,
          "infoEscalated": 111
        },
        "gpt-5.6-luna": {
          "mae": 3.3533333333333335,
          "signedError": 3.12,
          "overstated": 202,
          "infoCount": 150,
          "infoEscalated": 118
        }
      }
    },
    {
      "name": "Authentication",
      "n": 271,
      "models": {
        "jev": {
          "mae": 3.2715867158671585,
          "signedError": 3.2140221402214024,
          "overstated": 228,
          "infoCount": 49,
          "infoEscalated": 49
        },
        "haiku": {
          "mae": 3.1686346863468637,
          "signedError": 3.1398523985239852,
          "overstated": 215,
          "infoCount": 49,
          "infoEscalated": 49
        },
        "sonnet": {
          "mae": 1.7985239852398522,
          "signedError": 1.5402214022140222,
          "overstated": 151,
          "infoCount": 49,
          "infoEscalated": 49
        },
        "opus": {
          "mae": 1.7232472324723247,
          "signedError": 1.3911439114391144,
          "overstated": 146,
          "infoCount": 49,
          "infoEscalated": 47
        },
        "gpt-6-astra": {
          "mae": 1.6918819188191883,
          "signedError": 1.2062730627306273,
          "overstated": 140,
          "infoCount": 49,
          "infoEscalated": 41
        },
        "gpt-5.6-sol": {
          "mae": 1.940221402214022,
          "signedError": 1.481918819188192,
          "overstated": 152,
          "infoCount": 49,
          "infoEscalated": 47
        },
        "gpt-5.6-terra": {
          "mae": 2.1317343173431738,
          "signedError": 1.8955719557195574,
          "overstated": 177,
          "infoCount": 49,
          "infoEscalated": 46
        },
        "gpt-5.6-luna": {
          "mae": 2.7428044280442805,
          "signedError": 2.669741697416974,
          "overstated": 201,
          "infoCount": 49,
          "infoEscalated": 47
        }
      }
    },
    {
      "name": "Injection",
      "n": 92,
      "models": {
        "jev": {
          "mae": 2.1684782608695654,
          "signedError": 1.9793478260869564,
          "overstated": 66,
          "infoCount": 17,
          "infoEscalated": 14
        },
        "haiku": {
          "mae": 2.4195652173913045,
          "signedError": 2.3086956521739133,
          "overstated": 66,
          "infoCount": 17,
          "infoEscalated": 17
        },
        "sonnet": {
          "mae": 1.7228260869565217,
          "signedError": 1.1315217391304346,
          "overstated": 48,
          "infoCount": 17,
          "infoEscalated": 16
        },
        "opus": {
          "mae": 1.683695652173913,
          "signedError": 0.6228260869565218,
          "overstated": 43,
          "infoCount": 17,
          "infoEscalated": 13
        },
        "gpt-6-astra": {
          "mae": 1.4489130434782609,
          "signedError": 0.30978260869565216,
          "overstated": 31,
          "infoCount": 17,
          "infoEscalated": 9
        },
        "gpt-5.6-sol": {
          "mae": 1.675,
          "signedError": 0.9249999999999999,
          "overstated": 45,
          "infoCount": 17,
          "infoEscalated": 12
        },
        "gpt-5.6-terra": {
          "mae": 1.4945652173913044,
          "signedError": 0.8163043478260869,
          "overstated": 45,
          "infoCount": 17,
          "infoEscalated": 11
        },
        "gpt-5.6-luna": {
          "mae": 1.8499999999999999,
          "signedError": 1.3043478260869565,
          "overstated": 55,
          "infoCount": 17,
          "infoEscalated": 13
        }
      }
    },
    {
      "name": "Design / business logic",
      "n": 87,
      "models": {
        "jev": {
          "mae": 2.5,
          "signedError": 2.203448275862069,
          "overstated": 59,
          "infoCount": 24,
          "infoEscalated": 20
        },
        "haiku": {
          "mae": 2.8896551724137933,
          "signedError": 2.774712643678161,
          "overstated": 67,
          "infoCount": 24,
          "infoEscalated": 24
        },
        "sonnet": {
          "mae": 2.045977011494253,
          "signedError": 1.3954022988505745,
          "overstated": 46,
          "infoCount": 24,
          "infoEscalated": 24
        },
        "opus": {
          "mae": 1.9103448275862067,
          "signedError": 1.174712643678161,
          "overstated": 42,
          "infoCount": 24,
          "infoEscalated": 22
        },
        "gpt-6-astra": {
          "mae": 1.4954022988505746,
          "signedError": 0.07931034482758616,
          "overstated": 23,
          "infoCount": 24,
          "infoEscalated": 13
        },
        "gpt-5.6-sol": {
          "mae": 1.8287356321839079,
          "signedError": 0.8333333333333334,
          "overstated": 42,
          "infoCount": 24,
          "infoEscalated": 18
        },
        "gpt-5.6-terra": {
          "mae": 1.5999999999999999,
          "signedError": 0.8436781609195402,
          "overstated": 39,
          "infoCount": 24,
          "infoEscalated": 15
        },
        "gpt-5.6-luna": {
          "mae": 2.066666666666667,
          "signedError": 1.760919540229885,
          "overstated": 55,
          "infoCount": 24,
          "infoEscalated": 18
        }
      }
    },
    {
      "name": "Cryptography",
      "n": 69,
      "models": {
        "jev": {
          "mae": 4.153623188405797,
          "signedError": 3.7913043478260873,
          "overstated": 58,
          "infoCount": 23,
          "infoEscalated": 23
        },
        "haiku": {
          "mae": 4.030434782608696,
          "signedError": 3.955072463768116,
          "overstated": 56,
          "infoCount": 23,
          "infoEscalated": 23
        },
        "sonnet": {
          "mae": 2.527536231884058,
          "signedError": 2.1768115942028983,
          "overstated": 47,
          "infoCount": 23,
          "infoEscalated": 23
        },
        "opus": {
          "mae": 2.4507246376811596,
          "signedError": 2.0768115942028986,
          "overstated": 44,
          "infoCount": 23,
          "infoEscalated": 23
        },
        "gpt-6-astra": {
          "mae": 2.5869565217391304,
          "signedError": 1.6217391304347828,
          "overstated": 39,
          "infoCount": 23,
          "infoEscalated": 21
        },
        "gpt-5.6-sol": {
          "mae": 2.630434782608696,
          "signedError": 1.9782608695652173,
          "overstated": 42,
          "infoCount": 23,
          "infoEscalated": 22
        },
        "gpt-5.6-terra": {
          "mae": 2.681159420289855,
          "signedError": 1.866666666666667,
          "overstated": 44,
          "infoCount": 23,
          "infoEscalated": 21
        },
        "gpt-5.6-luna": {
          "mae": 3.723188405797101,
          "signedError": 3.427536231884058,
          "overstated": 56,
          "infoCount": 23,
          "infoEscalated": 22
        }
      }
    },
    {
      "name": "Resource consumption",
      "n": 35,
      "models": {
        "jev": {
          "mae": 3.2342857142857144,
          "signedError": 2.9714285714285715,
          "overstated": 25,
          "infoCount": 13,
          "infoEscalated": 12
        },
        "haiku": {
          "mae": 3.7685714285714287,
          "signedError": 3.7685714285714287,
          "overstated": 30,
          "infoCount": 13,
          "infoEscalated": 13
        },
        "sonnet": {
          "mae": 2.2657142857142856,
          "signedError": 2.1399999999999997,
          "overstated": 22,
          "infoCount": 13,
          "infoEscalated": 13
        },
        "opus": {
          "mae": 2.105714285714286,
          "signedError": 1.9285714285714286,
          "overstated": 20,
          "infoCount": 13,
          "infoEscalated": 12
        },
        "gpt-6-astra": {
          "mae": 1.9085714285714286,
          "signedError": 1.6057142857142859,
          "overstated": 17,
          "infoCount": 13,
          "infoEscalated": 9
        },
        "gpt-5.6-sol": {
          "mae": 2.1914285714285717,
          "signedError": 2.065714285714286,
          "overstated": 20,
          "infoCount": 13,
          "infoEscalated": 12
        },
        "gpt-5.6-terra": {
          "mae": 2.262857142857143,
          "signedError": 2.137142857142857,
          "overstated": 20,
          "infoCount": 13,
          "infoEscalated": 12
        },
        "gpt-5.6-luna": {
          "mae": 3.322857142857143,
          "signedError": 3.322857142857143,
          "overstated": 30,
          "infoCount": 13,
          "infoEscalated": 12
        }
      }
    },
    {
      "name": "Server-side request forgery",
      "n": 33,
      "models": {
        "jev": {
          "mae": 2.4939393939393937,
          "signedError": 1.6939393939393939,
          "overstated": 21,
          "infoCount": 5,
          "infoEscalated": 3
        },
        "haiku": {
          "mae": 2.596969696969697,
          "signedError": 2.4393939393939394,
          "overstated": 28,
          "infoCount": 5,
          "infoEscalated": 5
        },
        "sonnet": {
          "mae": 1.3484848484848484,
          "signedError": 1.2454545454545456,
          "overstated": 19,
          "infoCount": 5,
          "infoEscalated": 4
        },
        "opus": {
          "mae": 1.4454545454545455,
          "signedError": 1.3181818181818181,
          "overstated": 19,
          "infoCount": 5,
          "infoEscalated": 5
        },
        "gpt-6-astra": {
          "mae": 1.2272727272727273,
          "signedError": 0.2878787878787879,
          "overstated": 11,
          "infoCount": 5,
          "infoEscalated": 3
        },
        "gpt-5.6-sol": {
          "mae": 2.1363636363636362,
          "signedError": 1.8696969696969699,
          "overstated": 23,
          "infoCount": 5,
          "infoEscalated": 5
        },
        "gpt-5.6-terra": {
          "mae": 1.8,
          "signedError": 1.3393939393939396,
          "overstated": 22,
          "infoCount": 5,
          "infoEscalated": 5
        },
        "gpt-5.6-luna": {
          "mae": 2.0,
          "signedError": 1.7454545454545456,
          "overstated": 22,
          "infoCount": 5,
          "infoEscalated": 4
        }
      }
    },
    {
      "name": "AI / LLM security",
      "n": 27,
      "models": {
        "jev": {
          "mae": 3.6481481481481484,
          "signedError": 3.514814814814815,
          "overstated": 21,
          "infoCount": 14,
          "infoEscalated": 12
        },
        "haiku": {
          "mae": 4.737037037037037,
          "signedError": 4.737037037037037,
          "overstated": 26,
          "infoCount": 14,
          "infoEscalated": 14
        },
        "sonnet": {
          "mae": 3.281481481481481,
          "signedError": 2.9555555555555553,
          "overstated": 19,
          "infoCount": 14,
          "infoEscalated": 14
        },
        "opus": {
          "mae": 3.0518518518518523,
          "signedError": 2.7925925925925927,
          "overstated": 17,
          "infoCount": 14,
          "infoEscalated": 13
        },
        "gpt-6-astra": {
          "mae": 2.685185185185185,
          "signedError": 1.6185185185185185,
          "overstated": 14,
          "infoCount": 14,
          "infoEscalated": 12
        },
        "gpt-5.6-sol": {
          "mae": 2.8703703703703702,
          "signedError": 2.1148148148148147,
          "overstated": 15,
          "infoCount": 14,
          "infoEscalated": 13
        },
        "gpt-5.6-terra": {
          "mae": 2.874074074074074,
          "signedError": 2.2814814814814817,
          "overstated": 14,
          "infoCount": 14,
          "infoEscalated": 12
        },
        "gpt-5.6-luna": {
          "mae": 3.1185185185185187,
          "signedError": 2.755555555555555,
          "overstated": 16,
          "infoCount": 14,
          "infoEscalated": 12
        }
      }
    }
  ],
  "examples": [
    {
      "id": "csp",
      "referenceScore": 0.0,
      "referenceVector": "CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:N",
      "scores": {
        "gpt-5.6-luna": 0.0,
        "gpt-5.6-sol": 3.1,
        "gpt-5.6-terra": 0.0,
        "gpt-6-astra": 0.0,
        "haiku": 6.8,
        "jev": 7.5,
        "opus": 3.7,
        "sonnet": 3.4
      },
      "vectors": {
        "jev": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:H/I:N/A:N",
        "haiku": "CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:H/A:N",
        "sonnet": "CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:C/C:L/I:N/A:N",
        "opus": "CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:L/A:N",
        "gpt-6-astra": "CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:N/I:N/A:N",
        "gpt-5.6-sol": "CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:L/I:N/A:N",
        "gpt-5.6-terra": "CVSS:3.1/AV:N/AC:H/PR:N/UI:N/S:U/C:N/I:N/A:N",
        "gpt-5.6-luna": "CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:N/I:N/A:N"
      }
    },
    {
      "id": "oauth",
      "referenceScore": 0.0,
      "referenceVector": "CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:N/I:N/A:N",
      "scores": {
        "gpt-5.6-luna": 8.3,
        "gpt-5.6-sol": 6.8,
        "gpt-5.6-terra": 7.5,
        "gpt-6-astra": 6.8,
        "haiku": 8.1,
        "jev": 9.3,
        "opus": 5.3,
        "sonnet": 5.9
      },
      "vectors": {
        "jev": "CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:C/C:H/I:H/A:N",
        "haiku": "CVSS:3.1/AV:N/AC:L/PR:N/UI:R/S:U/C:H/I:H/A:N",
        "sonnet": "CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:L/A:N",
        "opus": "CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:N/A:N",
        "gpt-6-astra": "CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:H/A:N",
        "gpt-5.6-sol": "CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:H/A:N",
        "gpt-5.6-terra": "CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:U/C:H/I:H/A:H",
        "gpt-5.6-luna": "CVSS:3.1/AV:N/AC:H/PR:N/UI:R/S:C/C:H/I:H/A:H"
      }
    },
    {
      "id": "websocket",
      "referenceScore": 0.0,
      "referenceVector": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:N",
      "scores": {
        "gpt-5.6-luna": 7.5,
        "gpt-5.6-sol": 7.5,
        "gpt-5.6-terra": 5.3,
        "gpt-6-astra": 5.3,
        "haiku": 7.5,
        "jev": 7.5,
        "opus": 5.3,
        "sonnet": 5.3
      },
      "vectors": {
        "jev": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H",
        "haiku": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H",
        "sonnet": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L",
        "opus": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L",
        "gpt-6-astra": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L",
        "gpt-5.6-sol": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H",
        "gpt-5.6-terra": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:L",
        "gpt-5.6-luna": "CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:N/I:N/A:H"
      }
    },
    {
      "id": "instructions",
      "referenceScore": 0.0,
      "referenceVector": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:N/A:N",
      "scores": {
        "gpt-5.6-luna": 6.3,
        "gpt-5.6-sol": 4.3,
        "gpt-5.6-terra": 5.4,
        "gpt-6-astra": 4.3,
        "haiku": 8.1,
        "jev": 7.7,
        "opus": 4.3,
        "sonnet": 4.3
      },
      "vectors": {
        "jev": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:C/C:N/I:H/A:N",
        "haiku": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:N",
        "sonnet": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:N",
        "opus": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:N",
        "gpt-6-astra": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:N",
        "gpt-5.6-sol": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:N",
        "gpt-5.6-terra": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:N/I:L/A:L",
        "gpt-5.6-luna": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:L/I:L/A:L"
      }
    }
  ]
}
