{
  "status": "All attempts complete",
  "generated_at": "2026-10-09T02:56:02.986982+00:00",
  "completed_attempts": 10,
  "geometry_passes": 10,
  "counted_passes": 9,
  "planned_attempts": 10,
  "model_summary": {
    "astra": {
      "attempts": 5,
      "counted_attempts": 5,
      "geometry_valid_meshes": 5,
      "full_revisions": 10,
      "author_minutes_sum": 61.548383333333334,
      "usage_usd_sum": 15.06809,
      "parts": [
        {
          "ident": "t8_profile_01098d50",
          "counted": true,
          "geometry_valid": true,
          "full_revisions": 4,
          "author_minutes": 17.5566,
          "usage_usd": 4.657172,
          "usage_basis": "Standard API equivalent",
          "sj": 0.30546863311808253,
          "sj_2x": 0.30550220068861594,
          "blocks": 47,
          "nominal_hexes": 5086,
          "audit": {
            "initial_files_verified": 105,
            "actual_native_tool_evidence_archived": true,
            "manual_command_review_completed": true,
            "observed_deviations": [],
            "human_grade": null,
            "adapter_limitation": "Codex tool transport and reasoning effort differ from Claude Code; native requested model and local usage verified, provider snapshot/invoice unavailable.",
            "image_read_invocation_batches": 12,
            "visual_review_present": true,
            "supervisor_visual_observation": "Five-hole profile: 47-block layout with local level rings at lower/central/right holes; near-wall left hole fans to boundary; broad skewed transitions remain."
          }
        },
        {
          "ident": "t8_profile_011291e8",
          "counted": true,
          "geometry_valid": true,
          "full_revisions": 1,
          "author_minutes": 7.154533333333333,
          "usage_usd": 1.972924,
          "usage_basis": "Standard API equivalent",
          "sj": 0.4835406021956979,
          "sj_2x": 0.4835406021956939,
          "blocks": 110,
          "nominal_hexes": 2332,
          "audit": {
            "initial_files_verified": 105,
            "actual_native_tool_evidence_archived": true,
            "manual_command_review_completed": true,
            "observed_deviations": [],
            "human_grade": null,
            "adapter_limitation": "Codex tool transport and reasoning effort differ from Claude Code; native requested model and local usage verified, provider snapshot/invoice unavailable.",
            "image_read_invocation_batches": 4,
            "visual_review_present": true,
            "supervisor_visual_observation": "Four-loop rectilinear profile: 110-block layout; hole rings present, substantial cross-part subdivisions and diagonal fans by lower/right features."
          }
        },
        {
          "ident": "t8_block_0403c4d8",
          "counted": true,
          "geometry_valid": true,
          "full_revisions": 1,
          "author_minutes": 8.019833333333333,
          "usage_usd": 2.038754,
          "usage_basis": "Standard API equivalent",
          "sj": 0.3538920639486866,
          "sj_2x": 0.3538920639485729,
          "blocks": 65,
          "nominal_hexes": 11068,
          "audit": {
            "initial_files_verified": 105,
            "actual_native_tool_evidence_archived": true,
            "manual_command_review_completed": true,
            "observed_deviations": [],
            "human_grade": null,
            "adapter_limitation": "Codex tool transport and reasoning effort differ from Claude Code; native requested model and local usage verified, provider snapshot/invoice unavailable.",
            "image_read_invocation_batches": 5,
            "visual_review_present": true,
            "supervisor_visual_observation": "Stepped block: 65-block layout; sharp height-band splits retained; near-right hole has compact dense local transitions."
          }
        },
        {
          "ident": "t8_profile_01040740",
          "counted": true,
          "geometry_valid": true,
          "full_revisions": 1,
          "author_minutes": 10.855433333333334,
          "usage_usd": 2.591314,
          "usage_basis": "Standard API equivalent",
          "sj": 0.35565185718119247,
          "sj_2x": 0.3556518571811412,
          "blocks": 35,
          "nominal_hexes": 2244,
          "audit": {
            "initial_files_verified": 105,
            "actual_native_tool_evidence_archived": true,
            "manual_command_review_completed": true,
            "observed_deviations": [],
            "human_grade": null,
            "adapter_limitation": "Codex tool transport and reasoning effort differ from Claude Code; native requested model and local usage verified, provider snapshot/invoice unavailable.",
            "image_read_invocation_batches": 7,
            "visual_review_present": true
          }
        },
        {
          "ident": "t8_block_040566d0",
          "counted": true,
          "geometry_valid": true,
          "full_revisions": 3,
          "author_minutes": 17.961983333333333,
          "usage_usd": 3.807926,
          "usage_basis": "Standard API equivalent",
          "sj": 0.4292839429252336,
          "sj_2x": 0.42928394292522176,
          "blocks": 57,
          "nominal_hexes": 7874,
          "audit": {
            "initial_files_verified": 105,
            "actual_native_tool_evidence_archived": true,
            "manual_command_review_completed": true,
            "observed_deviations": [],
            "human_grade": null,
            "adapter_limitation": "Codex tool transport and reasoning effort differ from Claude Code; native requested model and local usage verified, provider snapshot/invoice unavailable.",
            "image_read_invocation_batches": 9,
            "visual_review_present": true
          }
        }
      ]
    },
    "opus": {
      "attempts": 5,
      "counted_attempts": 4,
      "geometry_valid_meshes": 5,
      "full_revisions": 10,
      "author_minutes_sum": 176.72,
      "usage_usd_sum": 31.9686588,
      "parts": [
        {
          "ident": "t8_profile_01098d50",
          "counted": true,
          "geometry_valid": true,
          "full_revisions": 5,
          "author_minutes": 52.68,
          "usage_usd": 9.535913000000003,
          "usage_basis": "Claude CLI reported usage",
          "sj": 0.566065678152422,
          "sj_2x": 0.5660656781524361,
          "blocks": 59,
          "nominal_hexes": 4484,
          "audit": {
            "tool_calls": 52,
            "tools": {
              "Read": 11,
              "Bash": 36,
              "Write": 5
            },
            "path_hits": 2,
            "hit_list": [
              [
                "a path outside the author's own",
                "Bash",
                "[local execution metadata omitted]"
              ],
              [
                "a search of the file tree",
                "Bash",
                "[local execution metadata omitted]"
              ]
            ],
            "checks_in_one_command": 0,
            "check_beside_a_loop": 0,
            "check_piped_into_a_filter": 0,
            "script_hits": [],
            "fast_checks_of_different_drafts": 0,
            "void_reasons": [],
            "supervisor_observed_deviations": [
              "Initial Read of witness.py first40lines and point() grep span exceeded point-only allowance. Frozen audit recorded path/listing warnings but void_reasons empty; no pinned/scorer change claimed."
            ]
          }
        },
        {
          "ident": "t8_profile_011291e8",
          "counted": true,
          "geometry_valid": true,
          "full_revisions": 1,
          "author_minutes": 22.6,
          "usage_usd": 4.5332122,
          "usage_basis": "Claude CLI reported usage",
          "sj": 0.4754275841254388,
          "sj_2x": 0.4754275841254428,
          "blocks": 57,
          "nominal_hexes": 1698,
          "audit": {
            "initial_files_verified": 105,
            "actual_native_tool_evidence_archived": true,
            "manual_command_review_completed": true,
            "observed_deviations": [
              "Read first 80 lines of witness.py outside permitted point() and grep point() with 45 following lines extending into read_rev. No pinned changes or other-model layout reads observed.",
              "Initial listing included own teacher-wave parent directory; original frozen audit recorded one path warning, no void reasons."
            ],
            "image_read_invocation_batches": 4,
            "visual_review_present": true,
            "human_grade": null,
            "geometry_acceptance": "Frozen independently reread counted true, stable true; supervisor deviations do not alter frozen acceptance.",
            "supervisor_visual_observations": [
              "Box-aligned rings around slots and pocket are visible. Circle has a compact polygonal ring with larger slanted background transitions; some elongated strips and acute transition corners remain. This is an observation, not an expert grade."
            ]
          }
        },
        {
          "ident": "t8_block_0403c4d8",
          "counted": false,
          "geometry_valid": true,
          "full_revisions": 2,
          "author_minutes": 64.97,
          "usage_usd": 10.1154298,
          "usage_basis": "Claude CLI reported usage",
          "sj": 0.42998793023082593,
          "sj_2x": 0.4299879302308103,
          "blocks": 58,
          "nominal_hexes": 13724,
          "audit": {
            "initial_files_verified": 105,
            "manual_command_review_completed": true,
            "actual_native_tool_evidence_archived": true,
            "image_read_invocation_batches": 5,
            "visual_review_present": true,
            "human_grade": null,
            "observed_deviations": [
              "Frozen audit void: 2 commands with checks of different drafts. Manual inspection finds each command echoes a quoted check command into cmd_times.txt then executes that same draft once; the text regex distinguishes quoted d1/d2 from unquoted names. calls.tsv confirms two single checks about four minutes apart, with full runs between. Apparent false positive retained; frozen audit is not modified.",
              "Initial own teacher-wave parent directory listing recorded as a path warning."
            ],
            "geometry_valid": true,
            "attempt_counted": false,
            "supervisor_visual_observations": [
              "Compact local rings surround holes, with larger irregular background transitions. Thin gaps and skewed transitions remain, consistent with author review. No human grade."
            ]
          }
        },
        {
          "ident": "t8_profile_01040740",
          "counted": true,
          "geometry_valid": true,
          "full_revisions": 1,
          "author_minutes": 25.53,
          "usage_usd": 5.4476977999999985,
          "usage_basis": "Claude CLI reported usage",
          "sj": 0.41671450486029854,
          "sj_2x": 0.4167145048602972,
          "blocks": 36,
          "nominal_hexes": 2152,
          "audit": {
            "initial_files_verified": 105,
            "manual_command_review_completed": true,
            "actual_native_tool_evidence_archived": true,
            "image_read_invocation_batches": 5,
            "visual_review_present": true,
            "observed_deviations": [
              "Own teacher-wave parent directory listing and recursive own-batch listing recorded as frozen audit warnings; no void reasons.",
              "Read witness.py first 5 lines and lines104\u2013153, extending beyond permitted point() into read_rev source."
            ],
            "human_grade": null
          }
        },
        {
          "ident": "t8_block_040566d0",
          "counted": true,
          "geometry_valid": true,
          "full_revisions": 1,
          "author_minutes": 10.94,
          "usage_usd": 2.336406,
          "usage_basis": "Claude CLI reported usage",
          "sj": 0.5464249070348385,
          "sj_2x": 0.5744945412948893,
          "blocks": 47,
          "nominal_hexes": 6902,
          "audit": {
            "initial_files_verified": 105,
            "manual_command_review_completed": true,
            "actual_native_tool_evidence_archived": true,
            "image_read_invocation_batches": 3,
            "visual_review_present": true,
            "observed_deviations": [
              "Own teacher-wave parent directory listing and recursive own-batch listing recorded as frozen audit warnings; no void reasons.",
              "Read point() plus 70 following witness.py lines, extending into read_rev/git_head/released source.",
              "Frozen audit check-piped warning appears in logged command text, not an additional executed draft check; no void reason."
            ],
            "human_grade": null
          }
        }
      ]
    }
  },
  "parts": [
    {
      "ident": "t8_profile_01098d50",
      "model": "GPT 6 Astra",
      "state": "complete",
      "counted": true,
      "geometry_valid": true,
      "attempt_failure_reasons": [],
      "blocks": 47,
      "full_iterations": 4,
      "minutes": 17.5566,
      "minimum_sj": 0.30546863311808253,
      "minimum_sj_2x": 0.30550220068861594,
      "usage_amount_usd": 4.657172,
      "usage_basis": "Standard API equivalent",
      "actual_subscription_charge_usd": null,
      "metrics": {
        "frame": {
          "origin": [
            0.0,
            0.0,
            1.00001e-07
          ],
          "u": [
            1.0,
            0.0,
            0.0
          ],
          "v": [
            0.0,
            1.0,
            0.0
          ]
        },
        "nominal": {
          "elements": 5086,
          "min_sj": 0.30546863311808253,
          "mean_sj": 0.8095375728345062,
          "p05_sj": 0.4650612948894153,
          "below_floor": 0,
          "inverted": 0
        },
        "doubled": {
          "elements": 40688,
          "min_sj": 0.30550220068861594,
          "mean_sj": 0.8352010327782684,
          "p05_sj": 0.49286240600392733,
          "below_floor": 0,
          "inverted": 0
        },
        "nominal_sha256": "d44a2da63b4fae00da4e33f169ab9c81558f082ba28bd08caad55352b1b23020",
        "doubled_sha256": "affb6b5fe312e087dfc2d62ba1355d9585d5bc02525f390b36f332389825810e",
        "source": "Original nominal VTU matched independent frozen reread; doubled VTU retained by replaying exact original x2.ron under pinned memory/slot guards. No model calls or acceptance changes."
      },
      "usage": {
        "token_usage": {
          "input_tokens": 2556880,
          "cached_input_tokens": 2452992,
          "cache_write_input_tokens": 0,
          "output_tokens": 23306,
          "reasoning_output_tokens": 6312,
          "total_tokens": 2580186
        },
        "model_configured": "gpt-6-astra",
        "model_resolved": null,
        "effort_observed": [
          "medium"
        ],
        "max_request_input_tokens": 121354,
        "standard_api_equivalent_usd": 4.657172,
        "fast_api_equivalent_usd": 9.314344,
        "actual_subscription_charge_usd": null,
        "pricing_source": "https://developers.openai.com/api/docs/models/gpt-6-astra",
        "standard_rates_usd_per_million": {
          "input": 10,
          "cached_input": 1,
          "output": 50
        },
        "method": "Final cumulative native usage, subtract cached input, output already includes reasoning. Actual service tier and provider snapshot unavailable. No accounting rerun."
      },
      "audit": {
        "initial_files_verified": 105,
        "actual_native_tool_evidence_archived": true,
        "manual_command_review_completed": true,
        "observed_deviations": [],
        "human_grade": null,
        "adapter_limitation": "Codex tool transport and reasoning effort differ from Claude Code; native requested model and local usage verified, provider snapshot/invoice unavailable.",
        "image_read_invocation_batches": 12,
        "visual_review_present": true,
        "supervisor_visual_observation": "Five-hole profile: 47-block layout with local level rings at lower/central/right holes; near-wall left hole fans to boundary; broad skewed transitions remain."
      },
      "human_grade": null
    },
    {
      "ident": "t8_profile_01098d50",
      "model": "Opus 5.5",
      "state": "complete",
      "counted": true,
      "geometry_valid": true,
      "attempt_failure_reasons": null,
      "blocks": 59,
      "full_iterations": 5,
      "minutes": 52.68,
      "minimum_sj": 0.566065678152422,
      "minimum_sj_2x": 0.5660656781524361,
      "usage_amount_usd": 9.535913000000003,
      "usage_basis": "Claude CLI reported usage",
      "actual_subscription_charge_usd": null,
      "metrics": {
        "frame": {
          "origin": [
            0.0,
            0.0,
            1.00001e-07
          ],
          "u": [
            1.0,
            0.0,
            0.0
          ],
          "v": [
            0.0,
            1.0,
            0.0
          ]
        },
        "nominal": {
          "elements": 4484,
          "min_sj": 0.566065678152422,
          "mean_sj": 0.9079001891202046,
          "p05_sj": 0.7073497101254058,
          "below_floor": 0,
          "inverted": 0
        },
        "doubled": {
          "elements": 35872,
          "min_sj": 0.5660656781524361,
          "mean_sj": 0.9266355917941849,
          "p05_sj": 0.7486373287660508,
          "below_floor": 0,
          "inverted": 0
        },
        "nominal_sha256": "dde46cdb59821d0b5963b90fbf49c85ea42758da1209eff2f0f4ee48b1f7e083",
        "doubled_sha256": "05eea7c3b3fe0dea3c7e0f2b35b130b141f934d3730e917f542223cc14a3f92f",
        "source": "Original nominal VTU matched independent frozen reread; doubled VTU replayed from exact original x2.ron under pinned memory/two shared slot guards. No model calls or acceptance changes."
      },
      "usage": null,
      "audit": {
        "tool_calls": 52,
        "tools": {
          "Read": 11,
          "Bash": 36,
          "Write": 5
        },
        "path_hits": 2,
        "hit_list": [
          [
            "a path outside the author's own",
            "Bash",
            "[local execution metadata omitted]"
          ],
          [
            "a search of the file tree",
            "Bash",
            "[local execution metadata omitted]"
          ]
        ],
        "checks_in_one_command": 0,
        "check_beside_a_loop": 0,
        "check_piped_into_a_filter": 0,
        "script_hits": [],
        "fast_checks_of_different_drafts": 0,
        "void_reasons": [],
        "supervisor_observed_deviations": [
          "Initial Read of witness.py first40lines and point() grep span exceeded point-only allowance. Frozen audit recorded path/listing warnings but void_reasons empty; no pinned/scorer change claimed."
        ]
      },
      "human_grade": null
    },
    {
      "ident": "t8_profile_011291e8",
      "model": "GPT 6 Astra",
      "state": "complete",
      "counted": true,
      "geometry_valid": true,
      "attempt_failure_reasons": [],
      "blocks": 110,
      "full_iterations": 1,
      "minutes": 7.154533333333333,
      "minimum_sj": 0.4835406021956979,
      "minimum_sj_2x": 0.4835406021956939,
      "usage_amount_usd": 1.972924,
      "usage_basis": "Standard API equivalent",
      "actual_subscription_charge_usd": null,
      "metrics": {
        "frame": {
          "origin": [
            -0.0,
            -0.0,
            1e-07
          ],
          "u": [
            1.0,
            0.0,
            0.0
          ],
          "v": [
            -0.0,
            1.0,
            0.0
          ]
        },
        "nominal": {
          "elements": 2332,
          "min_sj": 0.4835406021956979,
          "mean_sj": 0.9042187400792694,
          "p05_sj": 0.6801293187014773,
          "below_floor": 0,
          "inverted": 0
        },
        "doubled": {
          "elements": 18656,
          "min_sj": 0.4835406021956939,
          "mean_sj": 0.9205136511478914,
          "p05_sj": 0.705883087818915,
          "below_floor": 0,
          "inverted": 0
        },
        "nominal_sha256": "6357642f28ad2ad7c79ac94ad661aa4d90166923d31776c4d049d790c2e9ef58",
        "doubled_sha256": "ee15d3aba8b8e3ad43b5a67bad4210e81062737fab13909f4b5fb34b2d831d32",
        "source": "Original nominal VTU matched independent frozen reread; doubled VTU retained by replaying exact original x2.ron under pinned memory/slot guards. No model calls or acceptance changes."
      },
      "usage": {
        "token_usage": {
          "input_tokens": 1003141,
          "cached_input_tokens": 952704,
          "cache_write_input_tokens": 0,
          "output_tokens": 10317,
          "reasoning_output_tokens": 3681,
          "total_tokens": 1013458
        },
        "model_configured": "gpt-6-astra",
        "model_resolved": null,
        "effort_observed": [
          "medium"
        ],
        "max_request_input_tokens": 78433,
        "standard_api_equivalent_usd": 1.972924,
        "fast_api_equivalent_usd": 3.945848,
        "actual_subscription_charge_usd": null,
        "pricing_source": "https://developers.openai.com/api/docs/models/gpt-6-astra",
        "standard_rates_usd_per_million": {
          "input": 10,
          "cached_input": 1,
          "output": 50
        },
        "method": "Final cumulative native usage, subtract cached input, output already includes reasoning. Actual service tier and provider snapshot unavailable. No accounting rerun."
      },
      "audit": {
        "initial_files_verified": 105,
        "actual_native_tool_evidence_archived": true,
        "manual_command_review_completed": true,
        "observed_deviations": [],
        "human_grade": null,
        "adapter_limitation": "Codex tool transport and reasoning effort differ from Claude Code; native requested model and local usage verified, provider snapshot/invoice unavailable.",
        "image_read_invocation_batches": 4,
        "visual_review_present": true,
        "supervisor_visual_observation": "Four-loop rectilinear profile: 110-block layout; hole rings present, substantial cross-part subdivisions and diagonal fans by lower/right features."
      },
      "human_grade": null
    },
    {
      "ident": "t8_profile_011291e8",
      "model": "Opus 5.5",
      "state": "complete",
      "counted": true,
      "geometry_valid": true,
      "attempt_failure_reasons": null,
      "blocks": 57,
      "full_iterations": 1,
      "minutes": 22.6,
      "minimum_sj": 0.4754275841254388,
      "minimum_sj_2x": 0.4754275841254428,
      "usage_amount_usd": 4.5332122,
      "usage_basis": "Claude CLI reported usage",
      "actual_subscription_charge_usd": null,
      "metrics": {
        "frame": {
          "origin": [
            -0.0,
            -0.0,
            1e-07
          ],
          "u": [
            1.0,
            0.0,
            0.0
          ],
          "v": [
            -0.0,
            1.0,
            0.0
          ]
        },
        "nominal": {
          "elements": 1698,
          "min_sj": 0.4754275841254388,
          "mean_sj": 0.9047981134211486,
          "p05_sj": 0.7007198885373277,
          "below_floor": 0,
          "inverted": 0
        },
        "doubled": {
          "elements": 13584,
          "min_sj": 0.4754275841254428,
          "mean_sj": 0.9221894587283439,
          "p05_sj": 0.7196848968470252,
          "below_floor": 0,
          "inverted": 0
        },
        "nominal_sha256": "616faee3fff966ff319b84ad697149b5cbfaf1b8dfb901edb524547cb5e36661",
        "doubled_sha256": "4c3b526c85c37e880518885275bf3eec85895b2fe4431a25f018694c791db931",
        "source": "Original nominal VTU matched independent frozen reread; doubled VTU replayed from exact original x2.ron under pinned memory/two shared slot guards. No model calls or acceptance changes."
      },
      "usage": null,
      "audit": {
        "initial_files_verified": 105,
        "actual_native_tool_evidence_archived": true,
        "manual_command_review_completed": true,
        "observed_deviations": [
          "Read first 80 lines of witness.py outside permitted point() and grep point() with 45 following lines extending into read_rev. No pinned changes or other-model layout reads observed.",
          "Initial listing included own teacher-wave parent directory; original frozen audit recorded one path warning, no void reasons."
        ],
        "image_read_invocation_batches": 4,
        "visual_review_present": true,
        "human_grade": null,
        "geometry_acceptance": "Frozen independently reread counted true, stable true; supervisor deviations do not alter frozen acceptance.",
        "supervisor_visual_observations": [
          "Box-aligned rings around slots and pocket are visible. Circle has a compact polygonal ring with larger slanted background transitions; some elongated strips and acute transition corners remain. This is an observation, not an expert grade."
        ]
      },
      "human_grade": null
    },
    {
      "ident": "t8_block_0403c4d8",
      "model": "GPT 6 Astra",
      "state": "complete",
      "counted": true,
      "geometry_valid": true,
      "attempt_failure_reasons": [],
      "blocks": 65,
      "full_iterations": 1,
      "minutes": 8.019833333333333,
      "minimum_sj": 0.3538920639486866,
      "minimum_sj_2x": 0.3538920639485729,
      "usage_amount_usd": 2.038754,
      "usage_basis": "Standard API equivalent",
      "actual_subscription_charge_usd": null,
      "metrics": {
        "frame": {
          "origin": [
            0.0,
            0.0,
            1e-07
          ],
          "u": [
            1.0,
            0.0,
            0.0
          ],
          "v": [
            0.0,
            1.0,
            0.0
          ]
        },
        "nominal": {
          "elements": 11068,
          "min_sj": 0.3538920639486866,
          "mean_sj": 0.8525126680687699,
          "p05_sj": 0.536875452491431,
          "below_floor": 0,
          "inverted": 0
        },
        "doubled": {
          "elements": 88544,
          "min_sj": 0.3538920639485729,
          "mean_sj": 0.874492336338383,
          "p05_sj": 0.559424435170859,
          "below_floor": 0,
          "inverted": 0
        },
        "nominal_sha256": "dd0e8aca2fbfb45ef60f4d99a4c62f7abf02bb661acaca24894904af4052cec8",
        "doubled_sha256": "b5ae5aac082e56b826b1d645f969d639a297d843af0ab710e4d13c11a82a9c8e",
        "source": "Original nominal VTU matched independent frozen reread; doubled VTU retained by replaying exact original x2.ron under pinned memory/slot guards. No model calls or acceptance changes."
      },
      "usage": {
        "token_usage": {
          "input_tokens": 950631,
          "cached_input_tokens": 899584,
          "cache_write_input_tokens": 0,
          "output_tokens": 12574,
          "reasoning_output_tokens": 4177,
          "total_tokens": 963205
        },
        "model_configured": "gpt-6-astra",
        "model_resolved": null,
        "effort_observed": [
          "medium"
        ],
        "max_request_input_tokens": 79578,
        "standard_api_equivalent_usd": 2.038754,
        "fast_api_equivalent_usd": 4.077508,
        "actual_subscription_charge_usd": null,
        "pricing_source": "https://developers.openai.com/api/docs/models/gpt-6-astra",
        "standard_rates_usd_per_million": {
          "input": 10,
          "cached_input": 1,
          "output": 50
        },
        "method": "Final cumulative native usage, subtract cached input, output already includes reasoning. Actual service tier and provider snapshot unavailable. No accounting rerun."
      },
      "audit": {
        "initial_files_verified": 105,
        "actual_native_tool_evidence_archived": true,
        "manual_command_review_completed": true,
        "observed_deviations": [],
        "human_grade": null,
        "adapter_limitation": "Codex tool transport and reasoning effort differ from Claude Code; native requested model and local usage verified, provider snapshot/invoice unavailable.",
        "image_read_invocation_batches": 5,
        "visual_review_present": true,
        "supervisor_visual_observation": "Stepped block: 65-block layout; sharp height-band splits retained; near-right hole has compact dense local transitions."
      },
      "human_grade": null
    },
    {
      "ident": "t8_block_0403c4d8",
      "model": "Opus 5.5",
      "state": "complete",
      "counted": false,
      "geometry_valid": true,
      "attempt_failure_reasons": [
        "2 command(s) with checks of different drafts"
      ],
      "blocks": 58,
      "full_iterations": 2,
      "minutes": 64.97,
      "minimum_sj": 0.42998793023082593,
      "minimum_sj_2x": 0.4299879302308103,
      "usage_amount_usd": 10.1154298,
      "usage_basis": "Claude CLI reported usage",
      "actual_subscription_charge_usd": null,
      "metrics": {
        "frame": {
          "origin": [
            0.0,
            0.0,
            1e-07
          ],
          "u": [
            1.0,
            0.0,
            0.0
          ],
          "v": [
            0.0,
            1.0,
            0.0
          ]
        },
        "nominal": {
          "elements": 13724,
          "min_sj": 0.42998793023082593,
          "mean_sj": 0.8980280713077953,
          "p05_sj": 0.673743314690784,
          "below_floor": 0,
          "inverted": 0
        },
        "doubled": {
          "elements": 109792,
          "min_sj": 0.4299879302308103,
          "mean_sj": 0.9153810505051383,
          "p05_sj": 0.708277556715097,
          "below_floor": 0,
          "inverted": 0
        },
        "nominal_sha256": "fd12085584caa12dafe3bfc1fad9ae90f00c342f38f36657bed21833fc2e2e06",
        "doubled_sha256": "9bd981919e6ab06505bbc86c594f431fe90d3455a86f2c4fd5994f96ab4bddbf",
        "source": "Original nominal VTU matched independent frozen reread; doubled VTU replayed from exact original x2.ron under pinned memory/two shared slot guards. No model calls or acceptance changes."
      },
      "usage": null,
      "audit": {
        "initial_files_verified": 105,
        "manual_command_review_completed": true,
        "actual_native_tool_evidence_archived": true,
        "image_read_invocation_batches": 5,
        "visual_review_present": true,
        "human_grade": null,
        "observed_deviations": [
          "Frozen audit void: 2 commands with checks of different drafts. Manual inspection finds each command echoes a quoted check command into cmd_times.txt then executes that same draft once; the text regex distinguishes quoted d1/d2 from unquoted names. calls.tsv confirms two single checks about four minutes apart, with full runs between. Apparent false positive retained; frozen audit is not modified.",
          "Initial own teacher-wave parent directory listing recorded as a path warning."
        ],
        "geometry_valid": true,
        "attempt_counted": false,
        "supervisor_visual_observations": [
          "Compact local rings surround holes, with larger irregular background transitions. Thin gaps and skewed transitions remain, consistent with author review. No human grade."
        ]
      },
      "human_grade": null
    },
    {
      "ident": "t8_profile_01040740",
      "model": "GPT 6 Astra",
      "state": "complete",
      "counted": true,
      "geometry_valid": true,
      "attempt_failure_reasons": [],
      "blocks": 35,
      "full_iterations": 1,
      "minutes": 10.855433333333334,
      "minimum_sj": 0.35565185718119247,
      "minimum_sj_2x": 0.3556518571811412,
      "usage_amount_usd": 2.591314,
      "usage_basis": "Standard API equivalent",
      "actual_subscription_charge_usd": null,
      "metrics": {
        "frame": {
          "origin": [
            -0.0,
            -0.0,
            1e-07
          ],
          "u": [
            1.0,
            0.0,
            0.0
          ],
          "v": [
            -0.0,
            1.0,
            0.0
          ]
        },
        "nominal": {
          "elements": 2244,
          "min_sj": 0.35565185718119247,
          "mean_sj": 0.861704562641498,
          "p05_sj": 0.5387173683331066,
          "below_floor": 0,
          "inverted": 0
        },
        "doubled": {
          "elements": 17952,
          "min_sj": 0.3556518571811412,
          "mean_sj": 0.8825383346929545,
          "p05_sj": 0.5859845627021046,
          "below_floor": 0,
          "inverted": 0
        },
        "nominal_sha256": "1012c1e88e461d82ea5d36d0f615873f24f46a28331d381e5c511e9ab0916c6b",
        "doubled_sha256": "7ba21b622601c516d962027f4dac993036884683ec8f9a4cbf4104471c42b132",
        "source": "Original nominal VTU matched independent frozen reread; doubled VTU replayed from exact original x2.ron under pinned memory/two shared slot guards. No model calls or acceptance changes."
      },
      "usage": {
        "token_usage": {
          "input_tokens": 1206857,
          "cached_input_tokens": 1136384,
          "cache_write_input_tokens": 0,
          "output_tokens": 15004,
          "reasoning_output_tokens": 6221,
          "total_tokens": 1221861
        },
        "model_configured": "gpt-6-astra",
        "model_resolved": null,
        "effort_observed": [
          "medium"
        ],
        "max_request_input_tokens": 88250,
        "standard_api_equivalent_usd": 2.591314,
        "fast_api_equivalent_usd": 5.182628,
        "actual_subscription_charge_usd": null,
        "pricing_source": "https://developers.openai.com/api/docs/models/gpt-6-astra",
        "standard_rates_usd_per_million": {
          "input": 10,
          "cached_input": 1,
          "output": 50
        },
        "method": "Final cumulative native usage, subtract cached input, output already includes reasoning. Actual service tier and provider snapshot unavailable. No accounting rerun."
      },
      "audit": {
        "initial_files_verified": 105,
        "actual_native_tool_evidence_archived": true,
        "manual_command_review_completed": true,
        "observed_deviations": [],
        "human_grade": null,
        "adapter_limitation": "Codex tool transport and reasoning effort differ from Claude Code; native requested model and local usage verified, provider snapshot/invoice unavailable.",
        "image_read_invocation_batches": 7,
        "visual_review_present": true
      },
      "human_grade": null
    },
    {
      "ident": "t8_profile_01040740",
      "model": "Opus 5.5",
      "state": "complete",
      "counted": true,
      "geometry_valid": true,
      "attempt_failure_reasons": null,
      "blocks": 36,
      "full_iterations": 1,
      "minutes": 25.53,
      "minimum_sj": 0.41671450486029854,
      "minimum_sj_2x": 0.4167145048602972,
      "usage_amount_usd": 5.4476977999999985,
      "usage_basis": "Claude CLI reported usage",
      "actual_subscription_charge_usd": null,
      "metrics": {
        "frame": {
          "origin": [
            -0.0,
            -0.0,
            1e-07
          ],
          "u": [
            1.0,
            0.0,
            0.0
          ],
          "v": [
            -0.0,
            1.0,
            0.0
          ]
        },
        "nominal": {
          "elements": 2152,
          "min_sj": 0.41671450486029854,
          "mean_sj": 0.900878189061856,
          "p05_sj": 0.7071067811865296,
          "below_floor": 0,
          "inverted": 0
        },
        "doubled": {
          "elements": 17216,
          "min_sj": 0.4167145048602972,
          "mean_sj": 0.9182188419458822,
          "p05_sj": 0.7219882083561147,
          "below_floor": 0,
          "inverted": 0
        },
        "nominal_sha256": "f3864abcbc7a2415c8990825bc32319a1cf2def37ad19e0e8e337aa35852453e",
        "doubled_sha256": "ec567694162ca63e5d023924f1962015162cf22357c53761c78b3d6aae641cc5",
        "source": "Original nominal VTU matched independent frozen reread; doubled VTU replayed from exact original x2.ron under pinned memory/two shared slot guards. No model calls or acceptance changes."
      },
      "usage": null,
      "audit": {
        "initial_files_verified": 105,
        "manual_command_review_completed": true,
        "actual_native_tool_evidence_archived": true,
        "image_read_invocation_batches": 5,
        "visual_review_present": true,
        "observed_deviations": [
          "Own teacher-wave parent directory listing and recursive own-batch listing recorded as frozen audit warnings; no void reasons.",
          "Read witness.py first 5 lines and lines104\u2013153, extending beyond permitted point() into read_rev source."
        ],
        "human_grade": null
      },
      "human_grade": null
    },
    {
      "ident": "t8_block_040566d0",
      "model": "GPT 6 Astra",
      "state": "complete",
      "counted": true,
      "geometry_valid": true,
      "attempt_failure_reasons": [],
      "blocks": 57,
      "full_iterations": 3,
      "minutes": 17.961983333333333,
      "minimum_sj": 0.4292839429252336,
      "minimum_sj_2x": 0.42928394292522176,
      "usage_amount_usd": 3.807926,
      "usage_basis": "Standard API equivalent",
      "actual_subscription_charge_usd": null,
      "metrics": {
        "frame": {
          "origin": [
            -0.0,
            -0.0,
            1e-07
          ],
          "u": [
            1.0,
            0.0,
            0.0
          ],
          "v": [
            -0.0,
            1.0,
            0.0
          ]
        },
        "nominal": {
          "elements": 7874,
          "min_sj": 0.4292839429252336,
          "mean_sj": 0.9254616999499415,
          "p05_sj": 0.6346934229304251,
          "below_floor": 0,
          "inverted": 0
        },
        "doubled": {
          "elements": 62992,
          "min_sj": 0.42928394292522176,
          "mean_sj": 0.9421312173477282,
          "p05_sj": 0.7064585477950832,
          "below_floor": 0,
          "inverted": 0
        },
        "nominal_sha256": "05e352444e731bcfa56dbf079e268655d28da616900d97fdb646d78ced7deff7",
        "doubled_sha256": "ed7a8f8442f3228ccdc6f1aa0a234431a06c5cd38de1e2e0ab6e00f957ab6cc8",
        "source": "Original nominal VTU matched independent frozen reread; doubled VTU replayed from exact original x2.ron under pinned memory/two shared slot guards. No model calls or acceptance changes."
      },
      "usage": {
        "token_usage": {
          "input_tokens": 2126198,
          "cached_input_tokens": 2051456,
          "cache_write_input_tokens": 0,
          "output_tokens": 20181,
          "reasoning_output_tokens": 8948,
          "total_tokens": 2146379
        },
        "model_configured": "gpt-6-astra",
        "model_resolved": null,
        "effort_observed": [
          "medium"
        ],
        "max_request_input_tokens": 106478,
        "standard_api_equivalent_usd": 3.807926,
        "fast_api_equivalent_usd": 7.615852,
        "actual_subscription_charge_usd": null,
        "pricing_source": "https://developers.openai.com/api/docs/models/gpt-6-astra",
        "standard_rates_usd_per_million": {
          "input": 10,
          "cached_input": 1,
          "output": 50
        },
        "method": "Final cumulative native usage, subtract cached input, output already includes reasoning. Actual service tier and provider snapshot unavailable. No accounting rerun."
      },
      "audit": {
        "initial_files_verified": 105,
        "actual_native_tool_evidence_archived": true,
        "manual_command_review_completed": true,
        "observed_deviations": [],
        "human_grade": null,
        "adapter_limitation": "Codex tool transport and reasoning effort differ from Claude Code; native requested model and local usage verified, provider snapshot/invoice unavailable.",
        "image_read_invocation_batches": 9,
        "visual_review_present": true
      },
      "human_grade": null
    },
    {
      "ident": "t8_block_040566d0",
      "model": "Opus 5.5",
      "state": "complete",
      "counted": true,
      "geometry_valid": true,
      "attempt_failure_reasons": null,
      "blocks": 47,
      "full_iterations": 1,
      "minutes": 10.94,
      "minimum_sj": 0.5464249070348385,
      "minimum_sj_2x": 0.5744945412948893,
      "usage_amount_usd": 2.336406,
      "usage_basis": "Claude CLI reported usage",
      "actual_subscription_charge_usd": null,
      "metrics": {
        "frame": {
          "origin": [
            -0.0,
            -0.0,
            1e-07
          ],
          "u": [
            1.0,
            0.0,
            0.0
          ],
          "v": [
            -0.0,
            1.0,
            0.0
          ]
        },
        "nominal": {
          "elements": 6902,
          "min_sj": 0.5464249070348385,
          "mean_sj": 0.9444823368317831,
          "p05_sj": 0.7071067811865471,
          "below_floor": 0,
          "inverted": 0
        },
        "doubled": {
          "elements": 55216,
          "min_sj": 0.5744945412948893,
          "mean_sj": 0.9571967223215555,
          "p05_sj": 0.765298956849334,
          "below_floor": 0,
          "inverted": 0
        },
        "nominal_sha256": "810d5f1b1922fb372dd90fc806d13e67bde463412b59b107b23b7b120e609d2d",
        "doubled_sha256": "dc9fe228b9fbf4a16786aee92eb75c1c00d5f00acd3d1afb935f6c535d42032c",
        "source": "Original nominal VTU matched independent frozen reread; doubled VTU replayed from exact original x2.ron under pinned memory/two shared slot guards. No model calls or acceptance changes."
      },
      "usage": null,
      "audit": {
        "initial_files_verified": 105,
        "manual_command_review_completed": true,
        "actual_native_tool_evidence_archived": true,
        "image_read_invocation_batches": 3,
        "visual_review_present": true,
        "observed_deviations": [
          "Own teacher-wave parent directory listing and recursive own-batch listing recorded as frozen audit warnings; no void reasons.",
          "Read point() plus 70 following witness.py lines, extending into read_rev/git_head/released source.",
          "Frozen audit check-piped warning appears in logged command text, not an additional executed draft check; no void reason."
        ],
        "human_grade": null
      },
      "human_grade": null
    }
  ],
  "human_grades_assigned": false,
  "selection": "Five diverse difficult parts with verified Sonnet meshes, selected by many loops, height bands and blocks. No Sonnet layouts supplied to authors; a demonstrable feasible showcase sample, not an unbiased benchmark.",
  "session_time_cap_min": null,
  "session_spend_cap_usd": null
}