{
  "protocol": {
    "cases": [
      "cache",
      "ledger",
      "buildplan",
      "wallet"
    ],
    "seeds": [
      4242,
      8675309
    ],
    "temperature": 1.0,
    "top_p": 0.95,
    "top_k": 20,
    "output_limit": 8192,
    "seconds_limit": 480,
    "context": 131072,
    "runtime": "b11146",
    "opencode": "opencode v2.0.20",
    "model": "Qwen3.8-27B-Q4_K_M.gguf",
    "model_bytes": 16810714464,
    "draft_max": 3,
    "cache_prompt": false,
    "target_backend_sampling": false,
    "draft_backend_sampling": false,
    "server_restart": "before every task attempt",
    "workspace": "same absolute path for every attempt",
    "order": "per case: off/on for seed 4242, on/off for seed 8675309",
    "controls": "same committed fixture, prompt, working directory, agent title, tools, permissions and project config per pair",
    "request_audit": "abort on any difference in the canonical complete first chat-completion request",
    "scope": "two paired seeds per task; exploratory screen, not proof of general non-inferiority"
  },
  "pairs": [
    {
      "case": "cache",
      "seed": 4242,
      "identical_first_model_request": true,
      "identical_fixture": true,
      "only_server_argument_difference_is_mtp": true,
      "new_test_failures_with_mtp": [],
      "new_test_passes_with_mtp": [],
      "off": {
        "passed": 10,
        "tests": 10,
        "complete_and_passing": true,
        "seconds": 192.027,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": true,
        "tool_error_count": 0,
        "throughput": {
          "target_output_tokens": 5920,
          "generation_seconds": 155.98983,
          "generation_tokens_per_second": 37.95119207450896,
          "generation_requests": 6,
          "requests_reaching_output_limit": 0
        },
        "first_request_throughput": {
          "target_output_tokens": 110,
          "generation_seconds": 2.76131,
          "generation_tokens_per_second": 39.83616471891965,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 11090,
          "visible_characters": 2397,
          "completion_tokens": 5920,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "stop"
          ],
          "usage_requests": 6,
          "max_prompt_tokens": 9942,
          "all_reported_prefix_cache_counts_zero": true
        },
        "review": [
          {
            "probe": "finite_large_integer_ttl",
            "passed": true
          }
        ]
      },
      "on": {
        "passed": 10,
        "tests": 10,
        "complete_and_passing": true,
        "seconds": 153.523,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": true,
        "tool_error_count": 0,
        "throughput": {
          "target_output_tokens": 6082,
          "generation_seconds": 93.78266,
          "generation_tokens_per_second": 64.85207393349687,
          "generation_requests": 8,
          "requests_reaching_output_limit": 0
        },
        "first_request_throughput": {
          "target_output_tokens": 110,
          "generation_seconds": 1.3061800000000001,
          "generation_tokens_per_second": 84.21503927483195,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 9211,
          "visible_characters": 2748,
          "completion_tokens": 6082,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "stop"
          ],
          "usage_requests": 8,
          "max_prompt_tokens": 11035,
          "all_reported_prefix_cache_counts_zero": true
        },
        "review": [
          {
            "probe": "finite_large_integer_ttl",
            "passed": true
          }
        ]
      },
      "new_review_failures_with_mtp": [],
      "new_review_passes_with_mtp": []
    },
    {
      "case": "cache",
      "seed": 8675309,
      "identical_first_model_request": true,
      "identical_fixture": true,
      "only_server_argument_difference_is_mtp": true,
      "new_test_failures_with_mtp": [],
      "new_test_passes_with_mtp": [],
      "off": {
        "passed": 10,
        "tests": 10,
        "complete_and_passing": true,
        "seconds": 431.217,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": true,
        "tool_error_count": 1,
        "throughput": {
          "target_output_tokens": 11009,
          "generation_seconds": 306.96519,
          "generation_tokens_per_second": 35.86400138725828,
          "generation_requests": 11,
          "requests_reaching_output_limit": 0
        },
        "first_request_throughput": {
          "target_output_tokens": 109,
          "generation_seconds": 2.7771399999999997,
          "generation_tokens_per_second": 39.249011573057174,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 27666,
          "visible_characters": 3554,
          "completion_tokens": 11009,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "stop"
          ],
          "usage_requests": 11,
          "max_prompt_tokens": 18003,
          "all_reported_prefix_cache_counts_zero": true
        },
        "review": [
          {
            "probe": "finite_large_integer_ttl",
            "passed": true
          }
        ]
      },
      "on": {
        "passed": 10,
        "tests": 10,
        "complete_and_passing": true,
        "seconds": 259.004,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": true,
        "tool_error_count": 1,
        "throughput": {
          "target_output_tokens": 10260,
          "generation_seconds": 169.27704,
          "generation_tokens_per_second": 60.610700659699624,
          "generation_requests": 9,
          "requests_reaching_output_limit": 0
        },
        "first_request_throughput": {
          "target_output_tokens": 109,
          "generation_seconds": 1.4510699999999999,
          "generation_tokens_per_second": 75.1169826403964,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 25894,
          "visible_characters": 2688,
          "completion_tokens": 10260,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "stop"
          ],
          "usage_requests": 9,
          "max_prompt_tokens": 16555,
          "all_reported_prefix_cache_counts_zero": true
        },
        "review": [
          {
            "probe": "finite_large_integer_ttl",
            "passed": false
          }
        ]
      },
      "new_review_failures_with_mtp": [
        "finite_large_integer_ttl"
      ],
      "new_review_passes_with_mtp": []
    },
    {
      "case": "ledger",
      "seed": 4242,
      "identical_first_model_request": true,
      "identical_fixture": true,
      "only_server_argument_difference_is_mtp": true,
      "new_test_failures_with_mtp": [],
      "new_test_passes_with_mtp": [],
      "off": {
        "passed": 1,
        "tests": 10,
        "complete_and_passing": false,
        "seconds": 255.474,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": false,
        "tool_error_count": 0,
        "throughput": {
          "target_output_tokens": 8700,
          "generation_seconds": 240.04264999999998,
          "generation_tokens_per_second": 36.24355921749739,
          "generation_requests": 4,
          "requests_reaching_output_limit": 1
        },
        "first_request_throughput": {
          "target_output_tokens": 110,
          "generation_seconds": 2.81848,
          "generation_tokens_per_second": 39.02812863671198,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 32165,
          "visible_characters": 0,
          "completion_tokens": 8700,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "length"
          ],
          "usage_requests": 4,
          "max_prompt_tokens": 4823,
          "all_reported_prefix_cache_counts_zero": true
        },
        "review": [
          {
            "probe": "large_exact_amount",
            "passed": false
          },
          {
            "probe": "exact_header_names",
            "passed": false
          },
          {
            "probe": "read_time_io_error",
            "passed": false
          }
        ]
      },
      "on": {
        "passed": 1,
        "tests": 10,
        "complete_and_passing": false,
        "seconds": 178.766,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": false,
        "tool_error_count": 0,
        "throughput": {
          "target_output_tokens": 8700,
          "generation_seconds": 162.1489,
          "generation_tokens_per_second": 53.65438803470144,
          "generation_requests": 4,
          "requests_reaching_output_limit": 1
        },
        "first_request_throughput": {
          "target_output_tokens": 110,
          "generation_seconds": 1.3332899999999999,
          "generation_tokens_per_second": 82.50268133714347,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 31840,
          "visible_characters": 0,
          "completion_tokens": 8700,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "length"
          ],
          "usage_requests": 4,
          "max_prompt_tokens": 4823,
          "all_reported_prefix_cache_counts_zero": true
        },
        "review": [
          {
            "probe": "large_exact_amount",
            "passed": false
          },
          {
            "probe": "exact_header_names",
            "passed": false
          },
          {
            "probe": "read_time_io_error",
            "passed": false
          }
        ]
      },
      "new_review_failures_with_mtp": [],
      "new_review_passes_with_mtp": []
    },
    {
      "case": "ledger",
      "seed": 8675309,
      "identical_first_model_request": true,
      "identical_fixture": true,
      "only_server_argument_difference_is_mtp": true,
      "new_test_failures_with_mtp": [],
      "new_test_passes_with_mtp": [],
      "off": {
        "passed": 1,
        "tests": 10,
        "complete_and_passing": false,
        "seconds": 264.241,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": false,
        "tool_error_count": 1,
        "throughput": {
          "target_output_tokens": 8752,
          "generation_seconds": 242.71041,
          "generation_tokens_per_second": 36.05943395670585,
          "generation_requests": 5,
          "requests_reaching_output_limit": 1
        },
        "first_request_throughput": {
          "target_output_tokens": 109,
          "generation_seconds": 2.78065,
          "generation_tokens_per_second": 39.199467750346145,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 32145,
          "visible_characters": 0,
          "completion_tokens": 8752,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "length"
          ],
          "usage_requests": 5,
          "max_prompt_tokens": 5243,
          "all_reported_prefix_cache_counts_zero": true
        },
        "review": [
          {
            "probe": "large_exact_amount",
            "passed": false
          },
          {
            "probe": "exact_header_names",
            "passed": false
          },
          {
            "probe": "read_time_io_error",
            "passed": false
          }
        ]
      },
      "on": {
        "passed": 1,
        "tests": 10,
        "complete_and_passing": false,
        "seconds": 194.297,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": false,
        "tool_error_count": 1,
        "throughput": {
          "target_output_tokens": 8860,
          "generation_seconds": 169.20983,
          "generation_tokens_per_second": 52.361024179269016,
          "generation_requests": 5,
          "requests_reaching_output_limit": 1
        },
        "first_request_throughput": {
          "target_output_tokens": 109,
          "generation_seconds": 1.47226,
          "generation_tokens_per_second": 74.03583606156522,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 31762,
          "visible_characters": 0,
          "completion_tokens": 8860,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "length"
          ],
          "usage_requests": 5,
          "max_prompt_tokens": 7495,
          "all_reported_prefix_cache_counts_zero": true
        },
        "review": [
          {
            "probe": "large_exact_amount",
            "passed": false
          },
          {
            "probe": "exact_header_names",
            "passed": false
          },
          {
            "probe": "read_time_io_error",
            "passed": false
          }
        ]
      },
      "new_review_failures_with_mtp": [],
      "new_review_passes_with_mtp": []
    },
    {
      "case": "buildplan",
      "seed": 4242,
      "identical_first_model_request": true,
      "identical_fixture": true,
      "only_server_argument_difference_is_mtp": true,
      "new_test_failures_with_mtp": [],
      "new_test_passes_with_mtp": [],
      "off": {
        "passed": 1,
        "tests": 10,
        "complete_and_passing": false,
        "seconds": 256.33,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": false,
        "tool_error_count": 0,
        "throughput": {
          "target_output_tokens": 8700,
          "generation_seconds": 240.88231999999996,
          "generation_tokens_per_second": 36.117221056323274,
          "generation_requests": 4,
          "requests_reaching_output_limit": 1
        },
        "first_request_throughput": {
          "target_output_tokens": 110,
          "generation_seconds": 2.8193,
          "generation_tokens_per_second": 39.01677721420211,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 33053,
          "visible_characters": 0,
          "completion_tokens": 8700,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "length"
          ],
          "usage_requests": 4,
          "max_prompt_tokens": 4641,
          "all_reported_prefix_cache_counts_zero": true
        }
      },
      "on": {
        "passed": 1,
        "tests": 10,
        "complete_and_passing": false,
        "seconds": 168.852,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": false,
        "tool_error_count": 0,
        "throughput": {
          "target_output_tokens": 8700,
          "generation_seconds": 152.29135,
          "generation_tokens_per_second": 57.127341769575224,
          "generation_requests": 4,
          "requests_reaching_output_limit": 1
        },
        "first_request_throughput": {
          "target_output_tokens": 110,
          "generation_seconds": 1.3444800000000001,
          "generation_tokens_per_second": 81.81601808877781,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 33233,
          "visible_characters": 0,
          "completion_tokens": 8700,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "length"
          ],
          "usage_requests": 4,
          "max_prompt_tokens": 4641,
          "all_reported_prefix_cache_counts_zero": true
        }
      }
    },
    {
      "case": "buildplan",
      "seed": 8675309,
      "identical_first_model_request": true,
      "identical_fixture": true,
      "only_server_argument_difference_is_mtp": true,
      "new_test_failures_with_mtp": [],
      "new_test_passes_with_mtp": [],
      "off": {
        "passed": 1,
        "tests": 10,
        "complete_and_passing": false,
        "seconds": 275.924,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": false,
        "tool_error_count": 1,
        "throughput": {
          "target_output_tokens": 8904,
          "generation_seconds": 249.33214,
          "generation_tokens_per_second": 35.71140086472606,
          "generation_requests": 6,
          "requests_reaching_output_limit": 1
        },
        "first_request_throughput": {
          "target_output_tokens": 109,
          "generation_seconds": 2.80188,
          "generation_tokens_per_second": 38.902451211329534,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 32655,
          "visible_characters": 0,
          "completion_tokens": 8904,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "length"
          ],
          "usage_requests": 6,
          "max_prompt_tokens": 7033,
          "all_reported_prefix_cache_counts_zero": true
        }
      },
      "on": {
        "passed": 1,
        "tests": 10,
        "complete_and_passing": false,
        "seconds": 179.621,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": false,
        "tool_error_count": 1,
        "throughput": {
          "target_output_tokens": 8679,
          "generation_seconds": 163.12643,
          "generation_tokens_per_second": 53.20413129864976,
          "generation_requests": 4,
          "requests_reaching_output_limit": 1
        },
        "first_request_throughput": {
          "target_output_tokens": 109,
          "generation_seconds": 1.46741,
          "generation_tokens_per_second": 74.28053509244178,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 32829,
          "visible_characters": 0,
          "completion_tokens": 8679,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "length"
          ],
          "usage_requests": 4,
          "max_prompt_tokens": 4688,
          "all_reported_prefix_cache_counts_zero": true
        }
      }
    },
    {
      "case": "wallet",
      "seed": 4242,
      "identical_first_model_request": true,
      "identical_fixture": true,
      "only_server_argument_difference_is_mtp": true,
      "new_test_failures_with_mtp": [],
      "new_test_passes_with_mtp": [],
      "off": {
        "passed": 1,
        "tests": 10,
        "complete_and_passing": false,
        "seconds": 254.721,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": false,
        "tool_error_count": 0,
        "throughput": {
          "target_output_tokens": 8633,
          "generation_seconds": 239.05108,
          "generation_tokens_per_second": 36.11362057013087,
          "generation_requests": 4,
          "requests_reaching_output_limit": 1
        },
        "first_request_throughput": {
          "target_output_tokens": 110,
          "generation_seconds": 2.81759,
          "generation_tokens_per_second": 39.040456560393814,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 34740,
          "visible_characters": 0,
          "completion_tokens": 8633,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "length"
          ],
          "usage_requests": 4,
          "max_prompt_tokens": 4886,
          "all_reported_prefix_cache_counts_zero": true
        },
        "review": [
          {
            "probe": "destination_integer_overflow",
            "passed": false
          }
        ]
      },
      "on": {
        "passed": 1,
        "tests": 10,
        "complete_and_passing": false,
        "seconds": 161.89,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": false,
        "tool_error_count": 0,
        "throughput": {
          "target_output_tokens": 8633,
          "generation_seconds": 145.25869,
          "generation_tokens_per_second": 59.43190042537214,
          "generation_requests": 4,
          "requests_reaching_output_limit": 1
        },
        "first_request_throughput": {
          "target_output_tokens": 110,
          "generation_seconds": 1.32487,
          "generation_tokens_per_second": 83.02701397118207,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 35216,
          "visible_characters": 0,
          "completion_tokens": 8633,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "length"
          ],
          "usage_requests": 4,
          "max_prompt_tokens": 4886,
          "all_reported_prefix_cache_counts_zero": true
        },
        "review": [
          {
            "probe": "destination_integer_overflow",
            "passed": false
          }
        ]
      },
      "new_review_failures_with_mtp": [],
      "new_review_passes_with_mtp": []
    },
    {
      "case": "wallet",
      "seed": 8675309,
      "identical_first_model_request": true,
      "identical_fixture": true,
      "only_server_argument_difference_is_mtp": true,
      "new_test_failures_with_mtp": [
        "heldout.ContractTests.test_argument_validation_without_mutation",
        "heldout.ContractTests.test_concurrent_competing_debits",
        "heldout.ContractTests.test_concurrent_identical_retries",
        "heldout.ContractTests.test_conflicting_id_rejected_without_mutation",
        "heldout.ContractTests.test_credit_failure_rolls_back_debit",
        "heldout.ContractTests.test_insufficient_funds_and_reusable_id",
        "heldout.ContractTests.test_missing_accounts",
        "heldout.ContractTests.test_record_failure_rolls_back_both_balances",
        "heldout.ContractTests.test_replay_returns_original_snapshot"
      ],
      "new_test_passes_with_mtp": [],
      "off": {
        "passed": 10,
        "tests": 10,
        "complete_and_passing": false,
        "seconds": 480.114,
        "timed_out": true,
        "exit_code": -15,
        "protected_files_unchanged": true,
        "production_diff_present": true,
        "tool_error_count": 1,
        "throughput": {
          "target_output_tokens": 13214,
          "generation_seconds": 373.29143,
          "generation_tokens_per_second": 35.398616035733795,
          "generation_requests": 8,
          "requests_reaching_output_limit": 0
        },
        "first_request_throughput": {
          "target_output_tokens": 109,
          "generation_seconds": 2.78907,
          "generation_tokens_per_second": 39.0811274008899,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 34387,
          "visible_characters": 1473,
          "completion_tokens": 13214,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls"
          ],
          "usage_requests": 8,
          "max_prompt_tokens": 20057,
          "all_reported_prefix_cache_counts_zero": true
        },
        "review": [
          {
            "probe": "destination_integer_overflow",
            "passed": true
          }
        ]
      },
      "on": {
        "passed": 1,
        "tests": 10,
        "complete_and_passing": false,
        "seconds": 186.13,
        "timed_out": false,
        "exit_code": 0,
        "protected_files_unchanged": true,
        "production_diff_present": false,
        "tool_error_count": 1,
        "throughput": {
          "target_output_tokens": 8631,
          "generation_seconds": 164.27568,
          "generation_tokens_per_second": 52.5397307745127,
          "generation_requests": 5,
          "requests_reaching_output_limit": 1
        },
        "first_request_throughput": {
          "target_output_tokens": 109,
          "generation_seconds": 1.46386,
          "generation_tokens_per_second": 74.46067246867871,
          "scope": "same complete first request per pair; short tool-selection response"
        },
        "response": {
          "reasoning_characters": 36729,
          "visible_characters": 0,
          "completion_tokens": 8631,
          "finish_reasons": [
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "tool_calls",
            "length"
          ],
          "usage_requests": 5,
          "max_prompt_tokens": 5122,
          "all_reported_prefix_cache_counts_zero": true
        },
        "review": [
          {
            "probe": "destination_integer_overflow",
            "passed": false
          }
        ]
      },
      "new_review_failures_with_mtp": [
        "destination_integer_overflow"
      ],
      "new_review_passes_with_mtp": []
    }
  ],
  "completed_pairs": 8,
  "planned_pairs": 8,
  "limits": [
    "Two seeds per task; exploratory evidence, not statistical non-inferiority.",
    "Test methods within one task are correlated, not independent replications.",
    "Prefix caching is disabled and each attempt starts a fresh server; latency excludes server startup.",
    "Later model inputs necessarily diverge when the agent takes different actions.",
    "Generation throughput uses completed server timing records; an interrupted generation without a final timing is excluded.",
    "No simultaneous hardware workloads are intentionally added; desktop/background GPU activity is not fully controlled.",
    "Only accepted target output contributes to generation tokens/s, including reasoning and tool calls."
  ],
  "off": {
    "completed_tasks": 2,
    "attempts": 8,
    "passed_test_methods": 35,
    "test_methods": 80,
    "generation_tokens_per_second": 36.046116199658826,
    "target_output_tokens": 73832,
    "generation_seconds": 2048.26505,
    "median_task_elapsed_seconds": 260.28549999999996
  },
  "on": {
    "completed_tasks": 2,
    "attempts": 8,
    "passed_test_methods": 26,
    "test_methods": 80,
    "generation_tokens_per_second": 56.21342775056948,
    "target_output_tokens": 68545,
    "generation_seconds": 1219.37058,
    "median_task_elapsed_seconds": 179.1935
  },
  "paired_completion_regressions": [],
  "paired_completion_improvements": [],
  "paired_primary_test_regressions": [
    {
      "case": "wallet",
      "seed": 8675309,
      "off_passed": 10,
      "on_passed": 1,
      "new_failed_methods": [
        "heldout.ContractTests.test_argument_validation_without_mutation",
        "heldout.ContractTests.test_concurrent_competing_debits",
        "heldout.ContractTests.test_concurrent_identical_retries",
        "heldout.ContractTests.test_conflicting_id_rejected_without_mutation",
        "heldout.ContractTests.test_credit_failure_rolls_back_debit",
        "heldout.ContractTests.test_insufficient_funds_and_reusable_id",
        "heldout.ContractTests.test_missing_accounts",
        "heldout.ContractTests.test_record_failure_rolls_back_both_balances",
        "heldout.ContractTests.test_replay_returns_original_snapshot"
      ]
    }
  ],
  "paired_review_regressions": [
    {
      "case": "cache",
      "seed": 8675309,
      "probes": [
        "finite_large_integer_ttl"
      ]
    },
    {
      "case": "wallet",
      "seed": 8675309,
      "probes": [
        "destination_integer_overflow"
      ]
    }
  ],
  "successful_in_both_modes": [
    {
      "case": "cache",
      "seed": 4242,
      "off_seconds": 192.027,
      "on_seconds": 153.523,
      "mtp_elapsed_reduction_percent": 20.05134694600238
    },
    {
      "case": "cache",
      "seed": 8675309,
      "off_seconds": 431.217,
      "on_seconds": 259.004,
      "mtp_elapsed_reduction_percent": 39.93650528620161
    }
  ],
  "first_request_speed_ratios": [
    {
      "case": "cache",
      "seed": 4242,
      "ratio": 2.1140348190907834
    },
    {
      "case": "cache",
      "seed": 8675309,
      "ratio": 1.913856671283949
    },
    {
      "case": "ledger",
      "seed": 4242,
      "ratio": 2.1139287026828377
    },
    {
      "case": "ledger",
      "seed": 8675309,
      "ratio": 1.8886949316017552
    },
    {
      "case": "buildplan",
      "seed": 4242,
      "ratio": 2.096944543615375
    },
    {
      "case": "buildplan",
      "seed": 8675309,
      "ratio": 1.9094050060991814
    },
    {
      "case": "wallet",
      "seed": 4242,
      "ratio": 2.1266916754096625
    },
    {
      "case": "wallet",
      "seed": 8675309,
      "ratio": 1.9052846583689702
    }
  ]
}
