{
  "schema_version": 1,
  "kind": "live-browser-pinned-ab",
  "created_at": "2026-10-04T07:34:20.085751+00:00",
  "commits": {
    "baseline": "39d0bdc21dd3b17de452f413181f422a4c84af18",
    "treatment": "working-tree-snapshot"
  },
  "source_sha256": {
    "baseline": "329682d8b5e60d9ea81fe6a552ddc11af952c65f4b322db6f9bf42db77a3e18f",
    "treatment": "4a73f6d212df8f71a41348902558015627935bfd2f972e240ff2f69be8d449d0"
  },
  "repeats": 3,
  "tasks": [
    "already-satisfied-controls",
    "nested-scroll-target"
  ],
  "fixture_sha256": "3ad5004db9b7876bfc1dec5885207cdcc3611a33a8b38809804cfc16a7d3049a",
  "runtime": {
    "python": "3.14.2",
    "system": "Darwin"
  },
  "whole_benchmark_elapsed_ms": 27692.208,
  "timing_scope": "fixture setup/navigation + observe/decide/act + independent final checks; browser startup reported separately",
  "usage_scope": "actual provider-reported input/output tokens; failed or retried unknown usage is preserved",
  "pricing": {
    "usd_per_million_input": 0.042,
    "output_billed": false,
    "basis": "estimate using stated Jev price, not an invoice"
  },
  "scope": "synthetic loopback fixtures in new private Chromium processes; no customer/business execution",
  "summary": {
    "baseline": {
      "runs": 6,
      "passed": 3,
      "correct_completions": 3,
      "correct_gates": 0,
      "review_or_failure": 3,
      "statuses": {
        "done": 3,
        "blocked": 2,
        "needs_review": 1
      },
      "false_completions": 0,
      "wrong_actions": 0,
      "model_calls": 11,
      "provider_attempts": 11,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 16071,
        "output_tokens": 1131
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 16071,
          "output_tokens": 1131
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "whole_operation_total_ms": 8023.541,
      "whole_operation_median_ms": 1377.058,
      "request_bytes": 53177,
      "state_bytes": 5571,
      "estimated_input_usd": 0.000674982
    },
    "treatment": {
      "runs": 6,
      "passed": 3,
      "correct_completions": 3,
      "correct_gates": 0,
      "review_or_failure": 3,
      "statuses": {
        "done": 3,
        "blocked": 1,
        "needs_review": 2
      },
      "false_completions": 0,
      "wrong_actions": 0,
      "model_calls": 10,
      "provider_attempts": 10,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 14859,
        "output_tokens": 1005
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 14859,
          "output_tokens": 1005
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "whole_operation_total_ms": 7710.9,
      "whole_operation_median_ms": 1331.182,
      "request_bytes": 49651,
      "state_bytes": 5331,
      "estimated_input_usd": 0.000624078
    }
  },
  "runs": [
    {
      "task": "already-satisfied-controls",
      "runtime_version": "0.4.0",
      "expected": "done",
      "browser_startup_ms": 548.632,
      "passed": true,
      "correct_completion": true,
      "correct_gate": false,
      "review_or_failure": false,
      "false_completion": false,
      "wrong_actions": 0,
      "status": "done",
      "reason": null,
      "review_reasons": [],
      "pending_label": null,
      "error_type": null,
      "error": null,
      "steps": 1,
      "kernel_elapsed_ms": 1287,
      "kernel_usage_complete": true,
      "kernel_unknown_usage_attempts": 0,
      "independent_checks": {
        "confirmed": true,
        "exact_controls": true
      },
      "whole_operation_ms": 1377.682,
      "model_calls": 2,
      "provider_attempts": 2,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 3732,
        "output_tokens": 246
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 3732,
          "output_tokens": 246
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "request_bytes": 12600,
      "state_bytes": 1459,
      "estimated_input_usd": 0.000156744,
      "known_input_usd_lower_bound": 0.000156744,
      "calls": [
        {
          "model": "jev-1.13.0",
          "request_bytes": 6260,
          "state_bytes": 687,
          "usage": {
            "input_tokens": 1852,
            "output_tokens": 123
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 885.545
        },
        {
          "model": "jev-1.13.0",
          "request_bytes": 6340,
          "state_bytes": 772,
          "usage": {
            "input_tokens": 1880,
            "output_tokens": 123
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 354.496
        }
      ],
      "arm": "baseline",
      "repeat": 1
    },
    {
      "task": "already-satisfied-controls",
      "runtime_version": "0.4.0",
      "expected": "done",
      "browser_startup_ms": 665.093,
      "passed": true,
      "correct_completion": true,
      "correct_gate": false,
      "review_or_failure": false,
      "false_completion": false,
      "wrong_actions": 0,
      "status": "done",
      "reason": null,
      "review_reasons": [],
      "pending_label": null,
      "error_type": null,
      "error": null,
      "steps": 1,
      "kernel_elapsed_ms": 1320,
      "kernel_usage_complete": true,
      "kernel_unknown_usage_attempts": 0,
      "independent_checks": {
        "confirmed": true,
        "exact_controls": true
      },
      "whole_operation_ms": 1383.535,
      "model_calls": 2,
      "provider_attempts": 2,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 3662,
        "output_tokens": 232
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 3662,
          "output_tokens": 232
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "request_bytes": 12472,
      "state_bytes": 1475,
      "estimated_input_usd": 0.000153804,
      "known_input_usd_lower_bound": 0.000153804,
      "calls": [
        {
          "model": "jev-1.13.0",
          "request_bytes": 6196,
          "state_bytes": 695,
          "usage": {
            "input_tokens": 1817,
            "output_tokens": 116
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 923.895
        },
        {
          "model": "jev-1.13.0",
          "request_bytes": 6276,
          "state_bytes": 780,
          "usage": {
            "input_tokens": 1845,
            "output_tokens": 116
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 357.647
        }
      ],
      "arm": "treatment",
      "repeat": 1
    },
    {
      "task": "nested-scroll-target",
      "runtime_version": "0.4.0",
      "expected": "done",
      "browser_startup_ms": 458.455,
      "passed": false,
      "correct_completion": false,
      "correct_gate": false,
      "review_or_failure": true,
      "false_completion": false,
      "wrong_actions": 0,
      "status": "blocked",
      "reason": "model_blocked",
      "review_reasons": [],
      "pending_label": null,
      "error_type": null,
      "error": null,
      "steps": 1,
      "kernel_elapsed_ms": 1548,
      "kernel_usage_complete": true,
      "kernel_unknown_usage_attempts": 0,
      "independent_checks": {
        "target_effect": false
      },
      "whole_operation_ms": 1630.579,
      "model_calls": 2,
      "provider_attempts": 2,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 1959,
        "output_tokens": 159
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 1959,
          "output_tokens": 159
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "request_bytes": 6173,
      "state_bytes": 494,
      "estimated_input_usd": 8.2278e-05,
      "known_input_usd_lower_bound": 8.2278e-05,
      "calls": [
        {
          "model": "jev-1.13.0",
          "request_bytes": 3031,
          "state_bytes": 206,
          "usage": {
            "input_tokens": 957,
            "output_tokens": 75
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 1082.643
        },
        {
          "model": "jev-1.13.0",
          "request_bytes": 3142,
          "state_bytes": 288,
          "usage": {
            "input_tokens": 1002,
            "output_tokens": 84
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 409.4
        }
      ],
      "arm": "baseline",
      "repeat": 1
    },
    {
      "task": "nested-scroll-target",
      "runtime_version": "0.4.0",
      "expected": "done",
      "browser_startup_ms": 541.958,
      "passed": false,
      "correct_completion": false,
      "correct_gate": false,
      "review_or_failure": true,
      "false_completion": false,
      "wrong_actions": 0,
      "status": "blocked",
      "reason": "model_blocked",
      "review_reasons": [],
      "pending_label": null,
      "error_type": null,
      "error": null,
      "steps": 1,
      "kernel_elapsed_ms": 1199,
      "kernel_usage_complete": true,
      "kernel_unknown_usage_attempts": 0,
      "independent_checks": {
        "target_effect": false
      },
      "whole_operation_ms": 1278.829,
      "model_calls": 2,
      "provider_attempts": 2,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 1959,
        "output_tokens": 159
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 1959,
          "output_tokens": 159
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "request_bytes": 6173,
      "state_bytes": 494,
      "estimated_input_usd": 8.2278e-05,
      "known_input_usd_lower_bound": 8.2278e-05,
      "calls": [
        {
          "model": "jev-1.13.0",
          "request_bytes": 3031,
          "state_bytes": 206,
          "usage": {
            "input_tokens": 957,
            "output_tokens": 75
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 771.628
        },
        {
          "model": "jev-1.13.0",
          "request_bytes": 3142,
          "state_bytes": 288,
          "usage": {
            "input_tokens": 1002,
            "output_tokens": 84
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 380.613
        }
      ],
      "arm": "treatment",
      "repeat": 1
    },
    {
      "task": "already-satisfied-controls",
      "runtime_version": "0.4.0",
      "expected": "done",
      "browser_startup_ms": 773.987,
      "passed": true,
      "correct_completion": true,
      "correct_gate": false,
      "review_or_failure": false,
      "false_completion": false,
      "wrong_actions": 0,
      "status": "done",
      "reason": null,
      "review_reasons": [],
      "pending_label": null,
      "error_type": null,
      "error": null,
      "steps": 1,
      "kernel_elapsed_ms": 1469,
      "kernel_usage_complete": true,
      "kernel_unknown_usage_attempts": 0,
      "independent_checks": {
        "confirmed": true,
        "exact_controls": true
      },
      "whole_operation_ms": 1559.315,
      "model_calls": 2,
      "provider_attempts": 2,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 3662,
        "output_tokens": 232
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 3662,
          "output_tokens": 232
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "request_bytes": 12472,
      "state_bytes": 1475,
      "estimated_input_usd": 0.000153804,
      "known_input_usd_lower_bound": 0.000153804,
      "calls": [
        {
          "model": "jev-1.13.0",
          "request_bytes": 6196,
          "state_bytes": 695,
          "usage": {
            "input_tokens": 1817,
            "output_tokens": 116
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 1002.174
        },
        {
          "model": "jev-1.13.0",
          "request_bytes": 6276,
          "state_bytes": 780,
          "usage": {
            "input_tokens": 1845,
            "output_tokens": 116
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 399.385
        }
      ],
      "arm": "treatment",
      "repeat": 2
    },
    {
      "task": "already-satisfied-controls",
      "runtime_version": "0.4.0",
      "expected": "done",
      "browser_startup_ms": 442.747,
      "passed": true,
      "correct_completion": true,
      "correct_gate": false,
      "review_or_failure": false,
      "false_completion": false,
      "wrong_actions": 0,
      "status": "done",
      "reason": null,
      "review_reasons": [],
      "pending_label": null,
      "error_type": null,
      "error": null,
      "steps": 1,
      "kernel_elapsed_ms": 1273,
      "kernel_usage_complete": true,
      "kernel_unknown_usage_attempts": 0,
      "independent_checks": {
        "confirmed": true,
        "exact_controls": true
      },
      "whole_operation_ms": 1353.295,
      "model_calls": 2,
      "provider_attempts": 2,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 3732,
        "output_tokens": 246
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 3732,
          "output_tokens": 246
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "request_bytes": 12600,
      "state_bytes": 1459,
      "estimated_input_usd": 0.000156744,
      "known_input_usd_lower_bound": 0.000156744,
      "calls": [
        {
          "model": "jev-1.13.0",
          "request_bytes": 6260,
          "state_bytes": 687,
          "usage": {
            "input_tokens": 1852,
            "output_tokens": 123
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 858.267
        },
        {
          "model": "jev-1.13.0",
          "request_bytes": 6340,
          "state_bytes": 772,
          "usage": {
            "input_tokens": 1880,
            "output_tokens": 123
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 356.411
        }
      ],
      "arm": "baseline",
      "repeat": 2
    },
    {
      "task": "nested-scroll-target",
      "runtime_version": "0.4.0",
      "expected": "done",
      "browser_startup_ms": 404.866,
      "passed": false,
      "correct_completion": false,
      "correct_gate": false,
      "review_or_failure": true,
      "false_completion": false,
      "wrong_actions": 0,
      "status": "needs_review",
      "reason": "uncertain_decision",
      "review_reasons": [
        {
          "question": "operation",
          "reason": "below_min_top_probability",
          "actual": 0.53,
          "threshold": 0.55
        }
      ],
      "pending_label": "Scroll down in Results panel",
      "error_type": null,
      "error": null,
      "steps": 0,
      "kernel_elapsed_ms": 850,
      "kernel_usage_complete": true,
      "kernel_unknown_usage_attempts": 0,
      "independent_checks": {
        "target_effect": false
      },
      "whole_operation_ms": 927.132,
      "model_calls": 1,
      "provider_attempts": 1,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 957,
        "output_tokens": 75
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 957,
          "output_tokens": 75
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "request_bytes": 3031,
      "state_bytes": 206,
      "estimated_input_usd": 4.0194e-05,
      "known_input_usd_lower_bound": 4.0194e-05,
      "calls": [
        {
          "model": "jev-1.13.0",
          "request_bytes": 3031,
          "state_bytes": 206,
          "usage": {
            "input_tokens": 957,
            "output_tokens": 75
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 845.886
        }
      ],
      "arm": "treatment",
      "repeat": 2
    },
    {
      "task": "nested-scroll-target",
      "runtime_version": "0.4.0",
      "expected": "done",
      "browser_startup_ms": 379.737,
      "passed": false,
      "correct_completion": false,
      "correct_gate": false,
      "review_or_failure": true,
      "false_completion": false,
      "wrong_actions": 0,
      "status": "needs_review",
      "reason": "uncertain_decision",
      "review_reasons": [
        {
          "question": "operation",
          "reason": "below_min_top_probability",
          "actual": 0.51,
          "threshold": 0.55
        },
        {
          "question": "operation",
          "reason": "below_min_margin",
          "actual": 0.09000000000000002,
          "threshold": 0.1
        }
      ],
      "pending_label": "Scroll down in Results panel",
      "error_type": null,
      "error": null,
      "steps": 0,
      "kernel_elapsed_ms": 809,
      "kernel_usage_complete": true,
      "kernel_unknown_usage_attempts": 0,
      "independent_checks": {
        "target_effect": false
      },
      "whole_operation_ms": 889.779,
      "model_calls": 1,
      "provider_attempts": 1,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 957,
        "output_tokens": 75
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 957,
          "output_tokens": 75
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "request_bytes": 3031,
      "state_bytes": 206,
      "estimated_input_usd": 4.0194e-05,
      "known_input_usd_lower_bound": 4.0194e-05,
      "calls": [
        {
          "model": "jev-1.13.0",
          "request_bytes": 3031,
          "state_bytes": 206,
          "usage": {
            "input_tokens": 957,
            "output_tokens": 75
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 805.907
        }
      ],
      "arm": "baseline",
      "repeat": 2
    },
    {
      "task": "already-satisfied-controls",
      "runtime_version": "0.4.0",
      "expected": "done",
      "browser_startup_ms": 376.957,
      "passed": true,
      "correct_completion": true,
      "correct_gate": false,
      "review_or_failure": false,
      "false_completion": false,
      "wrong_actions": 0,
      "status": "done",
      "reason": null,
      "review_reasons": [],
      "pending_label": null,
      "error_type": null,
      "error": null,
      "steps": 1,
      "kernel_elapsed_ms": 1308,
      "kernel_usage_complete": true,
      "kernel_unknown_usage_attempts": 0,
      "independent_checks": {
        "confirmed": true,
        "exact_controls": true
      },
      "whole_operation_ms": 1395.772,
      "model_calls": 2,
      "provider_attempts": 2,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 3732,
        "output_tokens": 246
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 3732,
          "output_tokens": 246
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "request_bytes": 12600,
      "state_bytes": 1459,
      "estimated_input_usd": 0.000156744,
      "known_input_usd_lower_bound": 0.000156744,
      "calls": [
        {
          "model": "jev-1.13.0",
          "request_bytes": 6260,
          "state_bytes": 687,
          "usage": {
            "input_tokens": 1852,
            "output_tokens": 123
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 894.551
        },
        {
          "model": "jev-1.13.0",
          "request_bytes": 6340,
          "state_bytes": 772,
          "usage": {
            "input_tokens": 1880,
            "output_tokens": 123
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 362.056
        }
      ],
      "arm": "baseline",
      "repeat": 3
    },
    {
      "task": "already-satisfied-controls",
      "runtime_version": "0.4.0",
      "expected": "done",
      "browser_startup_ms": 463.084,
      "passed": true,
      "correct_completion": true,
      "correct_gate": false,
      "review_or_failure": false,
      "false_completion": false,
      "wrong_actions": 0,
      "status": "done",
      "reason": null,
      "review_reasons": [],
      "pending_label": null,
      "error_type": null,
      "error": null,
      "steps": 1,
      "kernel_elapsed_ms": 1412,
      "kernel_usage_complete": true,
      "kernel_unknown_usage_attempts": 0,
      "independent_checks": {
        "confirmed": true,
        "exact_controls": true
      },
      "whole_operation_ms": 1493.263,
      "model_calls": 2,
      "provider_attempts": 2,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 3662,
        "output_tokens": 232
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 3662,
          "output_tokens": 232
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "request_bytes": 12472,
      "state_bytes": 1475,
      "estimated_input_usd": 0.000153804,
      "known_input_usd_lower_bound": 0.000153804,
      "calls": [
        {
          "model": "jev-1.13.0",
          "request_bytes": 6196,
          "state_bytes": 695,
          "usage": {
            "input_tokens": 1817,
            "output_tokens": 116
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 992.767
        },
        {
          "model": "jev-1.13.0",
          "request_bytes": 6276,
          "state_bytes": 780,
          "usage": {
            "input_tokens": 1845,
            "output_tokens": 116
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 368.269
        }
      ],
      "arm": "treatment",
      "repeat": 3
    },
    {
      "task": "nested-scroll-target",
      "runtime_version": "0.4.0",
      "expected": "done",
      "browser_startup_ms": 454.145,
      "passed": false,
      "correct_completion": false,
      "correct_gate": false,
      "review_or_failure": true,
      "false_completion": false,
      "wrong_actions": 0,
      "status": "blocked",
      "reason": "model_blocked",
      "review_reasons": [],
      "pending_label": null,
      "error_type": null,
      "error": null,
      "steps": 1,
      "kernel_elapsed_ms": 1288,
      "kernel_usage_complete": true,
      "kernel_unknown_usage_attempts": 0,
      "independent_checks": {
        "target_effect": false
      },
      "whole_operation_ms": 1376.434,
      "model_calls": 2,
      "provider_attempts": 2,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 1959,
        "output_tokens": 159
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 1959,
          "output_tokens": 159
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "request_bytes": 6173,
      "state_bytes": 494,
      "estimated_input_usd": 8.2278e-05,
      "known_input_usd_lower_bound": 8.2278e-05,
      "calls": [
        {
          "model": "jev-1.13.0",
          "request_bytes": 3031,
          "state_bytes": 206,
          "usage": {
            "input_tokens": 957,
            "output_tokens": 75
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 873.085
        },
        {
          "model": "jev-1.13.0",
          "request_bytes": 3142,
          "state_bytes": 288,
          "usage": {
            "input_tokens": 1002,
            "output_tokens": 84
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 365.571
        }
      ],
      "arm": "baseline",
      "repeat": 3
    },
    {
      "task": "nested-scroll-target",
      "runtime_version": "0.4.0",
      "expected": "done",
      "browser_startup_ms": 400.625,
      "passed": false,
      "correct_completion": false,
      "correct_gate": false,
      "review_or_failure": true,
      "false_completion": false,
      "wrong_actions": 0,
      "status": "needs_review",
      "reason": "uncertain_decision",
      "review_reasons": [
        {
          "question": "operation",
          "reason": "below_min_top_probability",
          "actual": 0.52,
          "threshold": 0.55
        }
      ],
      "pending_label": "Scroll down in Results panel",
      "error_type": null,
      "error": null,
      "steps": 0,
      "kernel_elapsed_ms": 990,
      "kernel_usage_complete": true,
      "kernel_unknown_usage_attempts": 0,
      "independent_checks": {
        "target_effect": false
      },
      "whole_operation_ms": 1068.826,
      "model_calls": 1,
      "provider_attempts": 1,
      "provider_retries": 0,
      "failed_provider_calls": 0,
      "known_usage": {
        "input_tokens": 957,
        "output_tokens": 75
      },
      "model_usage": {
        "jev-1.13.0": {
          "input_tokens": 957,
          "output_tokens": 75
        }
      },
      "usage_complete": true,
      "unknown_usage_attempts": 0,
      "request_bytes": 3031,
      "state_bytes": 206,
      "estimated_input_usd": 4.0194e-05,
      "known_input_usd_lower_bound": 4.0194e-05,
      "calls": [
        {
          "model": "jev-1.13.0",
          "request_bytes": 3031,
          "state_bytes": 206,
          "usage": {
            "input_tokens": 957,
            "output_tokens": 75
          },
          "usage_complete": true,
          "unknown_usage_attempts": 0,
          "attempts": 1,
          "elapsed_ms": 986.503
        }
      ],
      "arm": "treatment",
      "repeat": 3
    }
  ],
  "limitations": [
    "Small fixture sets and repeated runs cannot establish production reliability or calibrated accuracy.",
    "This compares released program versions on the same Jev model, not Jev against an LLM.",
    "Needed confirmation is a correct expected stop; it is never resumed in the benchmark.",
    "Synthetic final-state checks reject a false DONE but do not prove every intermediate action was optimal."
  ]
}
