{
  "config": {
    "phase": "exploratory_followup",
    "models": [
      "qwen3.5:27b-q4_K_M",
      "gemma4:26b-a4b-it-qat"
    ],
    "methods": [
      "independent",
      "refine",
      "population",
      "archive"
    ],
    "instances_per_domain": 4,
    "dataset_seed": 20261108,
    "seeds": [
      17,
      43
    ],
    "token_budget": 4096,
    "max_output": 192,
    "min_output": 32,
    "temperature": 0.8,
    "num_ctx": 8192,
    "think": false,
    "schedule_seed": 20261108,
    "timeout_seconds": 300
  },
  "units": 192,
  "attempts": 1711,
  "physical_calls": 1423,
  "conditions": [
    {
      "model": "qwen3.5:27b-q4_K_M",
      "method": "independent",
      "units": 24,
      "quality": 0.8374692089424596,
      "mean_spent": 3873.125,
      "mean_unused": 222.875,
      "mean_attempts": 14.916666666666666,
      "feasible_rate": 1,
      "optimal_rate": 0.08333333333333333,
      "invalid_rate": 0.3100558659217877,
      "mean_seconds": 86.3810395098044
    },
    {
      "model": "qwen3.5:27b-q4_K_M",
      "method": "refine",
      "units": 24,
      "quality": 0.833081944576463,
      "mean_spent": 3888.2916666666665,
      "mean_unused": 207.70833333333334,
      "mean_attempts": 8.833333333333334,
      "feasible_rate": 0.9583333333333334,
      "optimal_rate": 0.08333333333333333,
      "invalid_rate": 0.32075471698113206,
      "mean_seconds": 67.34754152761889
    },
    {
      "model": "qwen3.5:27b-q4_K_M",
      "method": "population",
      "units": 24,
      "quality": 0.8629840045105692,
      "mean_spent": 3688.75,
      "mean_unused": 407.25,
      "mean_attempts": 7.208333333333333,
      "feasible_rate": 1,
      "optimal_rate": 0.08333333333333333,
      "invalid_rate": 0.23699421965317918,
      "mean_seconds": 56.09923023599549
    },
    {
      "model": "qwen3.5:27b-q4_K_M",
      "method": "archive",
      "units": 24,
      "quality": 0.8466967256635376,
      "mean_spent": 3381.875,
      "mean_unused": 714.125,
      "mean_attempts": 5,
      "feasible_rate": 1,
      "optimal_rate": 0.08333333333333333,
      "invalid_rate": 0.25833333333333336,
      "mean_seconds": 43.782760786448
    },
    {
      "model": "gemma4:26b-a4b-it-qat",
      "method": "independent",
      "units": 24,
      "quality": 0.9091615370593926,
      "mean_spent": 3865.9166666666665,
      "mean_unused": 230.08333333333334,
      "mean_attempts": 14.75,
      "feasible_rate": 1,
      "optimal_rate": 0.041666666666666664,
      "invalid_rate": 0.2033898305084746,
      "mean_seconds": 15.59623195996634
    },
    {
      "model": "gemma4:26b-a4b-it-qat",
      "method": "refine",
      "units": 24,
      "quality": 0.9270872987721145,
      "mean_spent": 3945.4583333333335,
      "mean_unused": 150.54166666666666,
      "mean_attempts": 8.666666666666666,
      "feasible_rate": 1,
      "optimal_rate": 0.08333333333333333,
      "invalid_rate": 0.2548076923076923,
      "mean_seconds": 12.12316440812962
    },
    {
      "model": "gemma4:26b-a4b-it-qat",
      "method": "population",
      "units": 24,
      "quality": 0.8735469062123348,
      "mean_spent": 3696,
      "mean_unused": 400,
      "mean_attempts": 7.041666666666667,
      "feasible_rate": 1,
      "optimal_rate": 0.08333333333333333,
      "invalid_rate": 0.21301775147928995,
      "mean_seconds": 10.007900459716135
    },
    {
      "model": "gemma4:26b-a4b-it-qat",
      "method": "archive",
      "units": 24,
      "quality": 0.8623726856275244,
      "mean_spent": 3354.6666666666665,
      "mean_unused": 741.3333333333334,
      "mean_attempts": 4.875,
      "feasible_rate": 1,
      "optimal_rate": 0,
      "invalid_rate": 0.26495726495726496,
      "mean_seconds": 7.813313690972184
    }
  ],
  "comparisons": [
    {
      "model": "qwen3.5:27b-q4_K_M",
      "comparison": "refine - independent",
      "paired_units": 24,
      "task_instances": 12,
      "delta": -0.004387264365996522,
      "ci95": [
        -0.032134920429866036,
        0.024275726722150007
      ],
      "wins": 9,
      "ties": 6,
      "losses": 9
    },
    {
      "model": "qwen3.5:27b-q4_K_M",
      "comparison": "population - independent",
      "paired_units": 24,
      "task_instances": 12,
      "delta": 0.02551479556810972,
      "ci95": [
        -0.0433274346034175,
        0.09068078117531823
      ],
      "wins": 8,
      "ties": 5,
      "losses": 11
    },
    {
      "model": "qwen3.5:27b-q4_K_M",
      "comparison": "archive - independent",
      "paired_units": 24,
      "task_instances": 12,
      "delta": 0.009227516721078091,
      "ci95": [
        -0.06262046694905678,
        0.08088343558721132
      ],
      "wins": 8,
      "ties": 6,
      "losses": 10
    },
    {
      "model": "qwen3.5:27b-q4_K_M",
      "comparison": "archive - population",
      "paired_units": 24,
      "task_instances": 12,
      "delta": -0.016287278847031628,
      "ci95": [
        -0.05129009697226373,
        0.015798182421402518
      ],
      "wins": 8,
      "ties": 7,
      "losses": 9
    },
    {
      "model": "gemma4:26b-a4b-it-qat",
      "comparison": "refine - independent",
      "paired_units": 24,
      "task_instances": 12,
      "delta": 0.017925761712721828,
      "ci95": [
        -0.004152150737227083,
        0.038376343565906114
      ],
      "wins": 13,
      "ties": 6,
      "losses": 5
    },
    {
      "model": "gemma4:26b-a4b-it-qat",
      "comparison": "population - independent",
      "paired_units": 24,
      "task_instances": 12,
      "delta": -0.03561463084705785,
      "ci95": [
        -0.060251537755877015,
        -0.013531991105400668
      ],
      "wins": 6,
      "ties": 4,
      "losses": 14
    },
    {
      "model": "gemma4:26b-a4b-it-qat",
      "comparison": "archive - independent",
      "paired_units": 24,
      "task_instances": 12,
      "delta": -0.04678885143186821,
      "ci95": [
        -0.06801922764041819,
        -0.02459818145948711
      ],
      "wins": 3,
      "ties": 6,
      "losses": 15
    },
    {
      "model": "gemma4:26b-a4b-it-qat",
      "comparison": "archive - population",
      "paired_units": 24,
      "task_instances": 12,
      "delta": -0.01117422058481036,
      "ci95": [
        -0.0337224833071044,
        0.009127397898756448
      ],
      "wins": 5,
      "ties": 10,
      "losses": 9
    }
  ]
}
