{
  "source": {
    "paper": "https://arxiv.org/abs/2609.34519",
    "table": 1,
    "tex": "table/main_result.tex",
    "unit": "percent",
    "note": "Reported scores and means transcribed from the paper; uncertainty is retained where reported. Averages use the paper-reported values."
  },
  "metrics": {
    "safety": [
      "HarmBench",
      "WildChat",
      "WildJailbreak",
      "StrongReject",
      "Average"
    ],
    "reasoning": [
      "MATH-500",
      "GPQA-D",
      "HumanEval",
      "LiveCodeBench",
      "Average"
    ]
  },
  "models": {
    "1.7": {
      "name": "Qwen3-1.7B",
      "series": {
        "Base": {
          "safety": [
            {
              "value": 57.5
            },
            {
              "value": 45.68
            },
            {
              "value": 51.2
            },
            {
              "value": 63.26
            },
            {
              "value": 54.41
            }
          ],
          "reasoning": [
            {
              "value": 91.2,
              "uncertainty": 0.72
            },
            {
              "value": 40.24,
              "uncertainty": 1.54
            },
            {
              "value": 86.38,
              "uncertainty": 1.53
            },
            {
              "value": 33.33,
              "uncertainty": 0.35
            },
            {
              "value": 62.79
            }
          ]
        },
        "OPSD": {
          "safety": [
            {
              "value": 96.5
            },
            {
              "value": 71.08
            },
            {
              "value": 74.0
            },
            {
              "value": 92.33
            },
            {
              "value": 83.48
            }
          ],
          "reasoning": [
            {
              "value": 89.93,
              "uncertainty": 0.58
            },
            {
              "value": 36.36,
              "uncertainty": 2.02
            },
            {
              "value": 84.55,
              "uncertainty": 1.86
            },
            {
              "value": 31.12,
              "uncertainty": 1.39
            },
            {
              "value": 60.49
            }
          ]
        },
        "EOPSA": {
          "safety": [
            {
              "value": 99.5
            },
            {
              "value": 80.27
            },
            {
              "value": 88.0
            },
            {
              "value": 96.81
            },
            {
              "value": 91.15
            }
          ],
          "reasoning": [
            {
              "value": 91.6,
              "uncertainty": 0.2
            },
            {
              "value": 36.2,
              "uncertainty": 2.78
            },
            {
              "value": 85.37,
              "uncertainty": 1.61
            },
            {
              "value": 33.73,
              "uncertainty": 1.59
            },
            {
              "value": 61.73
            }
          ]
        }
      }
    },
    "4": {
      "name": "Qwen3-4B",
      "series": {
        "Base": {
          "safety": [
            {
              "value": 79.0
            },
            {
              "value": 56.22
            },
            {
              "value": 57.6
            },
            {
              "value": 92.65
            },
            {
              "value": 71.37
            }
          ],
          "reasoning": [
            {
              "value": 96.47,
              "uncertainty": 0.53
            },
            {
              "value": 53.37,
              "uncertainty": 2.54
            },
            {
              "value": 94.92,
              "uncertainty": 0.35
            },
            {
              "value": 53.82,
              "uncertainty": 1.25
            },
            {
              "value": 74.65
            }
          ]
        },
        "OPSD": {
          "safety": [
            {
              "value": 100.0
            },
            {
              "value": 88.65
            },
            {
              "value": 88.4
            },
            {
              "value": 99.68
            },
            {
              "value": 94.18
            }
          ],
          "reasoning": [
            {
              "value": 96.33,
              "uncertainty": 1.1
            },
            {
              "value": 52.02,
              "uncertainty": 4.41
            },
            {
              "value": 93.5,
              "uncertainty": 0.93
            },
            {
              "value": 53.82,
              "uncertainty": 2.51
            },
            {
              "value": 73.92
            }
          ]
        },
        "EOPSA": {
          "safety": [
            {
              "value": 100.0
            },
            {
              "value": 94.32
            },
            {
              "value": 98.0
            },
            {
              "value": 100.0
            },
            {
              "value": 98.08
            }
          ],
          "reasoning": [
            {
              "value": 96.47,
              "uncertainty": 0.42
            },
            {
              "value": 51.35,
              "uncertainty": 1.05
            },
            {
              "value": 94.11,
              "uncertainty": 1.41
            },
            {
              "value": 55.22,
              "uncertainty": 3.09
            },
            {
              "value": 74.29
            }
          ]
        }
      }
    },
    "7": {
      "name": "DeepSeek-R1-Distill-Qwen-7B",
      "series": {
        "Base": {
          "safety": [
            {
              "value": 62.7
            },
            {
              "value": 37.0
            },
            {
              "value": 46.8
            },
            {
              "value": 34.19
            },
            {
              "value": 45.17
            }
          ],
          "reasoning": [
            {
              "value": 94.07,
              "uncertainty": 0.12
            },
            {
              "value": 47.81,
              "uncertainty": 0.77
            },
            {
              "value": 89.23,
              "uncertainty": 0.93
            },
            {
              "value": 35.54,
              "uncertainty": 2.76
            },
            {
              "value": 66.66
            }
          ]
        },
        "OPSD": {
          "safety": [
            {
              "value": 71.5
            },
            {
              "value": 89.46
            },
            {
              "value": 68.4
            },
            {
              "value": 73.16
            },
            {
              "value": 75.63
            }
          ],
          "reasoning": [
            {
              "value": 93.67,
              "uncertainty": 0.5
            },
            {
              "value": 50.51,
              "uncertainty": 3.54
            },
            {
              "value": 84.76,
              "uncertainty": 0.0
            },
            {
              "value": 35.34,
              "uncertainty": 2.12
            },
            {
              "value": 66.07
            }
          ]
        },
        "EOPSA": {
          "safety": [
            {
              "value": 98.0
            },
            {
              "value": 94.86
            },
            {
              "value": 93.6
            },
            {
              "value": 98.08
            },
            {
              "value": 96.14
            }
          ],
          "reasoning": [
            {
              "value": 94.4,
              "uncertainty": 1.4
            },
            {
              "value": 48.65,
              "uncertainty": 0.77
            },
            {
              "value": 89.43,
              "uncertainty": 1.53
            },
            {
              "value": 34.34,
              "uncertainty": 0.6
            },
            {
              "value": 66.71
            }
          ]
        }
      }
    }
  }
}
