.79 · 报告快照

来源：reports/20261112-huanxi-0.8.14.79-framework-tools-t1/comparisons/latest.md

# 框架版本效果对比

生成于 2026-11-12T08:20:17.085682+00:00；只读已完成试次，计划不计作结果。

| 产品/版本 | 日期 | 用户turn完成/总数 | 内容复核 | 回复终态合计/秒 | 模型响应 | 工具调用 | 参数/结果完全缺失 |
|---|---|---:|---|---:|---:|---:|---:|
| codex 0.165.0 | 2026-11-07 | 14/14 | PASS=9, INCONCLUSIVE=1 | 629.48 | 68 | 72 | 0/0 |
| openclaw 2026.11.1 | 2026-11-07 | 14/14 | PASS=9, INCONCLUSIVE=1 | 767.63 | 125 | 120 | 0/0 |
| huanxi 0.8.14.67 | 2026-11-07 | 11/14 | PASS=9, INCONCLUSIVE=1 | 3019.01 | 301 | 281 | 23/23 |
| huanxi 0.8.14.68 | 2026-11-08 | 14/14 | PASS=9, INCONCLUSIVE=1 | 1060.38 | 105 | 100 | 0/0 |
| huanxi 0.8.14.69 | 2026-11-08 | 14/14 | PASS=8, FAIL=1, INCONCLUSIVE=1 | 927.92 | 97 | 85 | 0/0 |
| huanxi 0.8.14.69 | 2026-11-12 | 14/14 | PASS=9, INCONCLUSIVE=1 | 804.15 | 85 | 74 | 0/0 |
| huanxi 0.8.14.79 | 2026-11-12 | 14/14 | PASS=9, INCONCLUSIVE=1 | 868.41 | 85 | 76 | 0/0 |
| huanxi 0.8.14.78 | 2026-11-12 | 14/14 | PASS=9, INCONCLUSIVE=1 | 902.85 | 85 | 72 | 0/0 |
| huanxi 0.8.14.77 | 2026-11-12 | 14/14 | PASS=9, INCONCLUSIVE=1 | 1018.20 | 98 | 88 | 0/0 |
| huanxi 0.8.14.80 | 2026-11-12 | 14/14 | PASS=9, INCONCLUSIVE=1 | 873.53 | 85 | 72 | 0/0 |
| huanxi 0.8.14.81 | 2026-11-12 | 14/14 | PASS=9, INCONCLUSIVE=1 | 948.27 | 90 | 77 | 0/0 |

## 工具正文覆盖

参数非空不等于全文完整；partial表示公开证据仅有部分视图，不能据此确定实际截断。unmarked表示旧采集未提供该标记，不能推断完整。原生分类与逐调用标记均保留。

| 产品/版本 | 参数availability | 结果availability |
|---|---|---|
| codex 0.165.0 | unmarked=72 | unmarked=72 |
| openclaw 2026.11.1 | unmarked=120 | unmarked=120 |
| huanxi 0.8.14.67 | available=258, unavailable=23 | available=258, unavailable=23 |
| huanxi 0.8.14.68 | available=93, partial=7 | available=100 |
| huanxi 0.8.14.69 | available=77, partial=8 | available=85 |
| huanxi 0.8.14.69 | available=68, partial=6 | available=74 |
| huanxi 0.8.14.79 | available=68, partial=8 | available=76 |
| huanxi 0.8.14.78 | available=60, partial=12 | available=71, partial=1 |
| huanxi 0.8.14.77 | available=75, partial=13 | available=88 |
| huanxi 0.8.14.80 | available=65, partial=7 | available=72 |
| huanxi 0.8.14.81 | available=71, partial=6 | available=77 |

## 准备、执行与收尾阶段

阶段表单列操作时间；正式试次阶段与用户回复计时重叠，不额外累加。缺测不补零。阶段PASS表示该操作完成，不代表所有题目内容通过。

| 产品/版本 | 阶段 | 秒 | 状态 |
|---|---|---:|---|
| codex 0.165.0 | 原归档未提供规范化阶段计时 | 缺测 | 未补测 |
| openclaw 2026.11.1 | 原归档未提供规范化阶段计时 | 缺测 | 未补测 |
| huanxi 0.8.14.67 | 原归档未提供规范化阶段计时 | 缺测 | 未补测 |
| huanxi 0.8.14.68 | sandbox_creation | 1.06 | PASS |
| huanxi 0.8.14.68 | python_and_release_download_integrity | 10.46 | PASS |
| huanxi 0.8.14.68 | base_dependencies_installation | 24.00 | PASS |
| huanxi 0.8.14.68 | eval_source_download_hash_check_and_dependencies | 91.29 | PASS |
| huanxi 0.8.14.68 | official_product_prepare_install_configure_fixture | 52.54 | PASS |
| huanxi 0.8.14.68 | model_configuration | 缺测 | PASS |
| huanxi 0.8.14.68 | case_material_seeding | 缺测 | PASS |
| huanxi 0.8.14.68 | health_version_models_schema_tools_preflight | 2.92 | PASS |
| huanxi 0.8.14.68 | serial_product_trial_and_per_case_observation | 1100.03 | PASS |
| huanxi 0.8.14.68 | native_result_readback | 0.08 | FAIL_CAPTURE |
| huanxi 0.8.14.68 | native_result_readback_v2 | 1.73 | PASS |
| huanxi 0.8.14.68 | original_artifact_grader | 0.14 | PASS |
| huanxi 0.8.14.68 | archive_pack_and_credential_scan | 4.17 | PASS |
| huanxi 0.8.14.68 | archive_transfer_and_local_hash_readback | 缺测 | PASS |
| huanxi 0.8.14.68 | stop_owned_services | 2.60 | PASS |
| huanxi 0.8.14.68 | sandbox_delete | 2.20 | PASS |
| huanxi 0.8.14.68 | sandbox_independent_absence_verification | 0.50 | PASS |
| huanxi 0.8.14.69 | sandbox_creation | 0.92 | PASS |
| huanxi 0.8.14.69 | python_and_release_download_integrity | 24.23 | PASS |
| huanxi 0.8.14.69 | base_dependencies_installation | 79.00 | PASS |
| huanxi 0.8.14.69 | eval_source_download_hash_check_and_dependencies | 170.01 | PASS |
| huanxi 0.8.14.69 | operator_signed_wheel_selector_preflight_attempt | 0.33 | PREPARATION_ERROR_CORRECTED |
| huanxi 0.8.14.69 | official_product_prepare_install_configure_fixture | 52.90 | PASS |
| huanxi 0.8.14.69 | prepare | 2.81 | PASS |
| huanxi 0.8.14.69 | install | 44.96 | PASS |
| huanxi 0.8.14.69 | model_configuration | 0.00 | PASS |
| huanxi 0.8.14.69 | case_material_seeding | 0.00 | PASS |
| huanxi 0.8.14.69 | start | 4.54 | PASS |
| huanxi 0.8.14.69 | health_version_models_schema_tools_preflight | 2.42 | PASS |
| huanxi 0.8.14.69 | serial_product_trial_and_per_case_observation | 970.60 | PASS |
| huanxi 0.8.14.69 | native_result_readback | 0.09 | FAIL_CAPTURE |
| huanxi 0.8.14.69 | native_result_readback_v2 | 1.33 | PASS |
| huanxi 0.8.14.69 | original_artifact_grader | 0.11 | PASS |
| huanxi 0.8.14.69 | archive_pack_and_credential_scan | 3.10 | PASS |
| huanxi 0.8.14.69 | archive_transfer_and_local_hash_readback | 缺测 | PASS |
| huanxi 0.8.14.69 | slow_logical_model_call_public_detail_readback | 3.79 | PASS |
| huanxi 0.8.14.69 | stop_owned_services | 3.01 | PASS |
| huanxi 0.8.14.69 | sandbox_delete | 1.19 | PASS |
| huanxi 0.8.14.69 | sandbox_independent_absence_verification | 0.50 | PASS |
| huanxi 0.8.14.69 | sandbox_creation | 2.47 | PASS |
| huanxi 0.8.14.69 | release_python_download_integrity | 18.19 | PASS |
| huanxi 0.8.14.69 | eval_codeload_download_failed | 0.57 | FAIL_CAPTURE |
| huanxi 0.8.14.69 | frozen_eval_source_recovery | 19.09 | PASS |
| huanxi 0.8.14.69 | official_os_install | 54.21 | PASS |
| huanxi 0.8.14.69 | product_trial_and_observation | 838.19 | PASS |
| huanxi 0.8.14.69 | pagination_probe | 0.07 | PASS |
| huanxi 0.8.14.69 | native_result_readback_v2 | 1.36 | PASS |
| huanxi 0.8.14.69 | original_artifact_grader | 0.12 | PASS |
| huanxi 0.8.14.69 | public_api_identity_preflight | 8.00 | PASS |
| huanxi 0.8.14.69 | archive_and_credential_scan | 3.22 | PASS |
| huanxi 0.8.14.69 | stop_owned_services | 2.56 | PASS |
| huanxi 0.8.14.69 | sandbox_delete | 14.41 | PASS |
| huanxi 0.8.14.69 | sandbox_independent_absence_verification | 2.99 | PASS |
| huanxi 0.8.14.79 | sandbox_creation | 1.46 | PASS |
| huanxi 0.8.14.79 | release_python_download_integrity | 13.22 | PASS |
| huanxi 0.8.14.79 | eval_codeload_download_failed | 0.88 | FAIL_CAPTURE |
| huanxi 0.8.14.79 | frozen_eval_source_recovery | 29.65 | PASS |
| huanxi 0.8.14.79 | official_os_install | 66.03 | PASS |
| huanxi 0.8.14.79 | product_trial_and_observation | 903.24 | PASS |
| huanxi 0.8.14.79 | pagination_probe | 0.10 | PASS |
| huanxi 0.8.14.79 | native_result_readback_v2 | 1.42 | PASS |
| huanxi 0.8.14.79 | original_artifact_grader | 0.18 | PASS |
| huanxi 0.8.14.79 | public_api_identity_preflight | 2.68 | PASS |
| huanxi 0.8.14.79 | archive_and_credential_scan | 4.10 | PASS |
| huanxi 0.8.14.79 | stop_owned_services | 3.04 | PASS |
| huanxi 0.8.14.79 | sandbox_delete | 1.70 | PASS |
| huanxi 0.8.14.79 | sandbox_independent_absence_verification | 0.53 | PASS |
| huanxi 0.8.14.78 | sandbox_creation | 1.38 | PASS |
| huanxi 0.8.14.78 | release_python_download_integrity | 13.37 | PASS |
| huanxi 0.8.14.78 | eval_codeload_download_failed | 0.56 | FAIL_CAPTURE |
| huanxi 0.8.14.78 | frozen_eval_source_recovery | 18.74 | PASS |
| huanxi 0.8.14.78 | official_os_install | 65.12 | PASS |
| huanxi 0.8.14.78 | product_trial_and_observation | 938.20 | PASS |
| huanxi 0.8.14.78 | pagination_probe | 0.09 | PASS |
| huanxi 0.8.14.78 | native_result_readback_v2 | 1.53 | PASS |
| huanxi 0.8.14.78 | original_artifact_grader | 0.34 | PASS |
| huanxi 0.8.14.78 | public_api_identity_preflight | 3.08 | PASS |
| huanxi 0.8.14.78 | archive_and_credential_scan | 4.43 | PASS |
| huanxi 0.8.14.78 | stop_owned_services | 3.40 | PASS |
| huanxi 0.8.14.78 | sandbox_delete | 1.15 | PASS |
| huanxi 0.8.14.78 | sandbox_independent_absence_verification | 0.54 | PASS |
| huanxi 0.8.14.77 | sandbox_creation | 1.66 | PASS |
| huanxi 0.8.14.77 | release_python_download_integrity | 13.93 | PASS |
| huanxi 0.8.14.77 | eval_codeload_download_failed | 0.58 | FAIL_CAPTURE |
| huanxi 0.8.14.77 | frozen_eval_source_recovery | 21.48 | PASS |
| huanxi 0.8.14.77 | official_os_install | 65.15 | PASS |
| huanxi 0.8.14.77 | product_trial_and_observation | 1054.20 | PASS |
| huanxi 0.8.14.77 | pagination_probe | 0.11 | PASS |
| huanxi 0.8.14.77 | native_result_readback_v2 | 1.68 | PASS |
| huanxi 0.8.14.77 | original_artifact_grader | 0.15 | PASS |
| huanxi 0.8.14.77 | public_api_identity_preflight | 2.97 | PASS |
| huanxi 0.8.14.77 | archive_and_credential_scan | 4.29 | PASS |
| huanxi 0.8.14.77 | stop_owned_services | 4.75 | PASS |
| huanxi 0.8.14.77 | sandbox_delete | 3.20 | PASS |
| huanxi 0.8.14.77 | sandbox_independent_absence_verification | 1.20 | PASS |
| huanxi 0.8.14.80 | sandbox_creation | 1.46 | PASS |
| huanxi 0.8.14.80 | release_python_download_integrity | 14.32 | PASS |
| huanxi 0.8.14.80 | eval_codeload_download_failed | 0.59 | FAIL_CAPTURE |
| huanxi 0.8.14.80 | frozen_eval_source_recovery | 19.22 | PASS |
| huanxi 0.8.14.80 | official_os_install | 64.49 | PASS |
| huanxi 0.8.14.80 | product_trial_and_observation | 1539.76 | PASS |
| huanxi 0.8.14.80 | pagination_probe | 0.08 | PASS |
| huanxi 0.8.14.80 | native_result_readback_v2 | 1.19 | PASS |
| huanxi 0.8.14.80 | original_artifact_grader | 0.17 | PASS |
| huanxi 0.8.14.80 | public_api_identity_preflight | 3.19 | PASS |
| huanxi 0.8.14.80 | archive_and_credential_scan | 4.02 | PASS |
| huanxi 0.8.14.80 | stop_owned_services | 2.08 | PASS |
| huanxi 0.8.14.80 | sandbox_delete | 1.40 | PASS |
| huanxi 0.8.14.80 | sandbox_independent_absence_verification | 0.49 | PASS |
| huanxi 0.8.14.80 | c01_observation_contract_gap_and_recovery | 628.79 | RECOVERED |
| huanxi 0.8.14.81 | sandbox_creation | 1.89 | PASS |
| huanxi 0.8.14.81 | release_python_download_integrity | 15.65 | PASS |
| huanxi 0.8.14.81 | eval_codeload_download_failed | 0.58 | FAIL_CAPTURE |
| huanxi 0.8.14.81 | frozen_eval_source_recovery | 19.69 | PASS |
| huanxi 0.8.14.81 | official_os_install | 63.94 | PASS |
| huanxi 0.8.14.81 | product_trial_and_observation | 1617.05 | PASS |
| huanxi 0.8.14.81 | pagination_probe | 0.08 | PASS |
| huanxi 0.8.14.81 | native_result_readback_v2 | 1.54 | PASS |
| huanxi 0.8.14.81 | original_artifact_grader | 0.11 | PASS |
| huanxi 0.8.14.81 | public_api_identity_preflight | 2.57 | PASS |
| huanxi 0.8.14.81 | archive_and_credential_scan | 3.27 | PASS |
| huanxi 0.8.14.81 | stop_owned_services | 3.92 | PASS |
| huanxi 0.8.14.81 | sandbox_delete | 0.88 | PASS |
| huanxi 0.8.14.81 | sandbox_independent_absence_verification | 0.44 | PASS |
| huanxi 0.8.14.81 | c01_observation_contract_gap_and_recovery | 633.18 | RECOVERED |

## 任务范围与模型观察

范围计数为可见命令中明确出现 /tmp 的次数，独立于内容 PASS；不是完整系统调用审计。

| 产品/版本 | /tmp 命令观察数 | 涉及题目 | 有范围观察的题数 | 模型ID缺测/不符 |
|---|---:|---|---:|---:|
| codex 0.165.0 | 4 | C04, C10 | 10 | 0/0 |
| openclaw 2026.11.1 | 5 | C04, C09 | 10 | 0/0 |
| huanxi 0.8.14.67 | 34 | C01, C03, C04, C06, C07, C09, C10 | 10 | 0/0 |
| huanxi 0.8.14.68 | 15 | C01, C02, C03, C04, C07, C09, C10 | 10 | 0/0 |
| huanxi 0.8.14.69 | 12 | C02, C03, C08, C09 | 10 | 0/0 |
| huanxi 0.8.14.69 | 7 | C03, C09, C10 | 10 | 0/0 |
| huanxi 0.8.14.79 | 9 | C04, C09, C10 | 10 | 0/0 |
| huanxi 0.8.14.78 | 11 | C03, C04, C05, C09, C10 | 10 | 0/0 |
| huanxi 0.8.14.77 | 0 |  | 10 | 0/0 |
| huanxi 0.8.14.80 | 2 | C08, C09 | 10 | 0/0 |
| huanxi 0.8.14.81 | 1 | C10 | 10 | 0/0 |

## 每题耗时与步骤

每格：用户输入→回复/失败终态秒数之和；模型响应数/工具调用数；题级内容复核。

| Case | codex 0.165.0 | openclaw 2026.11.1 | huanxi 0.8.14.67 | huanxi 0.8.14.68 | huanxi 0.8.14.69 | huanxi 0.8.14.69 | huanxi 0.8.14.79 | huanxi 0.8.14.78 | huanxi 0.8.14.77 | huanxi 0.8.14.80 | huanxi 0.8.14.81 |
|---|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|---:|
| C01 发票去重与对账 | 30.61；4/3；PASS | 36.86；6/6；PASS | 173.55；18/14；PASS | 32.34；4/3；PASS | 57.88；7/6；PASS | 40.97；5/4；PASS | 38.02；5/4；PASS | 40.95；5/4；PASS | 46.83；4/3；PASS | 42.63；5/4；PASS | 37.73；4/3；PASS |
| C02 日志关联与事故时间线 | 46.81；4/8；PASS | 51.20；8/9；PASS | 306.24；35/34；PASS | 76.11；8/8；PASS | 72.57；10/9；PASS | 100.87；9/9；PASS | 68.78；5/5；PASS | 81.04；7/6；PASS | 103.27；10/10；PASS | 58.04；6/5；PASS | 80.84；7/6；PASS |
| C03 多文件政策优先级 | 29.13；5/6；PASS | 49.03；9/10；PASS | 134.09；17/18；PASS | 50.55；7/7；PASS | 38.49；6/5；PASS | 46.17；6/5；PASS | 29.86；4/3；PASS | 44.25；5/4；PASS | 62.83；7/6；PASS | 34.84；4/3；PASS | 53.38；6/5；PASS |
| C04 时区与稳定排序 | 52.27；8/9；PASS | 78.99；11/11；PASS | 384.39；38/34；PASS | 100.43；8/9；PASS | 39.75；5/4；PASS | 69.13；8/7；PASS | 66.04；6/5；PASS | 63.86；6/5；PASS | 51.71；5/5；PASS | 98.29；9/8；PASS | 55.72；6/5；PASS |
| C05 小模块修复与回归 | 92.96；7/10；PASS | 134.43；21/22；PASS | 145.86；12/11；PASS | 123.86；12/11；PASS | 127.05；9/8；PASS | 92.03；9/8；PASS | 78.29；8/7；PASS | 143.15；12/12；PASS | 112.15；9/8；PASS | 97.58；8/7；PASS | 158.17；14/13；PASS |
| C06 依赖图与关键路径 | 23.34；3/2；PASS | 55.14；9/8；PASS | 478.43；41/37；PASS | 81.20；8/7；PASS | 113.45；9/8；FAIL | 53.71；5/4；PASS | 56.94；5/4；PASS | 54.42；5/4；PASS | 61.40；7/6；PASS | 46.92；4/3；PASS | 47.24；5/4；PASS |
| C07 采购方案追加预算约束 | 73.86；6/5；PASS | 111.12；17/16；PASS | 114.22；10/8；PASS | 135.77；15/13；PASS | 85.21；8/6；PASS | 86.62；10/8；PASS | 90.00；10/8；PASS | 74.99；7/5；PASS | 93.38；8/8；PASS | 130.22；13/12；PASS | 139.03；14/12；PASS |
| C08 销售汇总与增量更正 | 78.90；9/11；PASS | 52.84；10/8；PASS | 197.36；21/16；PASS | 96.76；10/8；PASS | 77.06；12/11；PASS | 72.35；8/8；PASS | 93.23；9/7；PASS | 70.39；7/5；PASS | 105.19；11/9；PASS | 103.63；10/8；PASS | 98.64；8/6；PASS |
| C09 命令行工具连续迭代 | 153.57；15/13；PASS | 91.57；15/13；PASS | 510.96；50/48；PASS | 187.25；17/18；PASS | 217.11；18/16；PASS | 144.90；14/12；PASS | 181.92；21/21；PASS | 147.35；16/14；PASS | 262.12；26/24；PASS | 125.21；13/11；PASS | 160.60；15/13；PASS |
| C10 发布判断与新证据修正 | 48.03；7/5；INCONCLUSIVE | 106.46；19/17；INCONCLUSIVE | 573.91；59/61；INCONCLUSIVE | 176.13；16/16；INCONCLUSIVE | 99.34；13/12；INCONCLUSIVE | 97.41；11/9；INCONCLUSIVE | 165.33；12/12；INCONCLUSIVE | 182.45；15/13；INCONCLUSIVE | 119.31；11/9；INCONCLUSIVE | 136.18；13/11；INCONCLUSIVE | 116.92；11/10；INCONCLUSIVE |

## OS版本变化（相邻已完成试次）

| 旧版→新版 | 回复合计差/秒 | 变化比例 | 完成turn数变化 | 模型/工具次数变化 |
|---|---:|---:|---:|---:|
| 0.8.14.67 → 0.8.14.68 | -1958.62 | -64.9% | +3 | -196/-181 |
| 0.8.14.68 → 0.8.14.69 | -132.46 | -12.5% | +0 | -8/-15 |
| 0.8.14.69 → 0.8.14.69 | -123.77 | -13.3% | +0 | -12/-11 |
| 0.8.14.69 → 0.8.14.79 | 64.26 | +8.0% | +0 | +0/+2 |
| 0.8.14.79 → 0.8.14.78 | 34.44 | +4.0% | +0 | +0/-4 |
| 0.8.14.78 → 0.8.14.77 | 115.35 | +12.8% | +0 | +13/+16 |
| 0.8.14.77 → 0.8.14.80 | -144.68 | -14.2% | +0 | -13/-16 |
| 0.8.14.80 → 0.8.14.81 | 74.74 | +8.6% | +0 | +5/+5 |

### 最新两次OS逐输入变化

负值表示本次观察更少/更快；终态变化必须一起看，失败更快不能当作性能提升。

| 输入 | 原终态→现终态 | 题级内容原→现 | 原耗时→现耗时/秒 | 耗时差/秒 | 模型/工具次数差 |
|---|---|---|---:|---:|---:|
| C01/U1 | completed → completed | PASS → PASS | 42.63 → 37.73 | -4.90 | -1/-1 |
| C02/U1 | completed → completed | PASS → PASS | 58.04 → 80.84 | 22.80 | 1/1 |
| C03/U1 | completed → completed | PASS → PASS | 34.84 → 53.38 | 18.54 | 2/2 |
| C04/U1 | completed → completed | PASS → PASS | 98.29 → 55.72 | -42.57 | -3/-3 |
| C05/U1 | completed → completed | PASS → PASS | 97.58 → 158.17 | 60.59 | 6/6 |
| C06/U1 | completed → completed | PASS → PASS | 46.92 → 47.24 | 0.32 | 1/1 |
| C07/U1 | completed → completed | PASS → PASS | 57.15 → 42.47 | -14.68 | -1/-1 |
| C07/U2 | completed → completed | PASS → PASS | 73.07 → 96.56 | 23.49 | 2/1 |
| C08/U1 | completed → completed | PASS → PASS | 33.76 → 39.86 | 6.10 | 0/0 |
| C08/U2 | completed → completed | PASS → PASS | 69.86 → 58.78 | -11.08 | -2/-2 |
| C09/U1 | completed → completed | PASS → PASS | 51.27 → 56.58 | 5.30 | 0/0 |
| C09/U2 | completed → completed | PASS → PASS | 73.94 → 104.03 | 30.09 | 2/2 |
| C10/U1 | completed → completed | INCONCLUSIVE → INCONCLUSIVE | 56.49 → 65.16 | 8.68 | 0/1 |
| C10/U2 | completed → completed | INCONCLUSIVE → INCONCLUSIVE | 79.69 → 51.76 | -27.94 | -2/-2 |

## 下一步优化定位

以下按最新已完成OS试次与同题最快固定基线的延迟差排序，仅用于选择需要查看的轨迹，不证明原因。内容FAIL需独立优先处理，不能以更快抵消。

| Case | 内容 | OS秒 | 固定基线最快秒 | 差值秒 | OS模型/工具 |
|---|---|---:|---:|---:|---:|
| C09 命令行工具连续迭代 | PASS | 160.60 | 91.57 | +69.04 | 15/13 |
| C05 小模块修复与回归 | PASS | 158.17 | 92.96 | +65.21 | 14/13 |
| C07 采购方案追加预算约束 | PASS | 139.03 | 73.86 | +65.17 | 14/12 |
| C08 销售汇总与增量更正 | PASS | 98.64 | 52.84 | +45.80 | 8/6 |
| C02 日志关联与事故时间线 | PASS | 80.84 | 46.81 | +34.02 | 7/6 |

判分歧义题不进入优化排名；各题原始延迟仍全部保留。优化分析应沿这些case的user-turns/model-steps/tool-steps查看：延迟来自更多轮次、单轮更慢、工具等待还是观察缺口；由原始日志支持后才能列为已证实原因。

## 证据与可比边界

- codex 0.165.0：run `20261107-codex-0.165.0-t1`；逐工具缺测与截断见工具CSV，完整采集说明见私有manifest。
- openclaw 2026.11.1：run `20261107-openclaw-2026.11.1-t1`；逐工具缺测与截断见工具CSV，完整采集说明见私有manifest。
- huanxi 0.8.14.67：run `20261107-huanxi-0.8.14.67-t1`；逐工具缺测与截断见工具CSV，完整采集说明见私有manifest。
- huanxi 0.8.14.68：run `20261108-huanxi-0.8.14.68-framework-tools-t1`；逐工具缺测与截断见工具CSV，完整采集说明见私有manifest。
- huanxi 0.8.14.69：run `20261108-huanxi-0.8.14.69-framework-tools-t1`；逐工具缺测与截断见工具CSV，完整采集说明见私有manifest。
- huanxi 0.8.14.69：run `20261112-huanxi-0.8.14.69-framework-tools-t1-v2env`；逐工具缺测与截断见工具CSV，完整采集说明见私有manifest。
- huanxi 0.8.14.79：run `20261112-huanxi-0.8.14.79-framework-tools-t1`；逐工具缺测与截断见工具CSV，完整采集说明见私有manifest。
- huanxi 0.8.14.78：run `20261112-huanxi-0.8.14.78-framework-tools-t1`；逐工具缺测与截断见工具CSV，完整采集说明见私有manifest。
- huanxi 0.8.14.77：run `20261112-huanxi-0.8.14.77-framework-tools-t1`；逐工具缺测与截断见工具CSV，完整采集说明见私有manifest。
- huanxi 0.8.14.80：run `20261112-huanxi-0.8.14.80-framework-tools-t1`；逐工具缺测与截断见工具CSV，完整采集说明见私有manifest。
- huanxi 0.8.14.81：run `20261112-huanxi-0.8.14.81-framework-tools-t1`；逐工具缺测与截断见工具CSV，完整采集说明见私有manifest。

固定基线：`20261107-codex-0.165.0-t1`, `20261107-openclaw-2026.11.1-t1`

历史复核：C08 原检查器漏查子目录备份，原判保留、内容复核 PASS；C10 部署清单完整性与 JSON 层级题面有歧义，保留 INCONCLUSIVE。
历史采集：Codex CLI 终态后等待累计约147.30秒，另列；OpenClaw C01 CLI退出时间缺测。OS .67 C08/U2由只读恢复拿到46.885秒真实回复延迟，与C09/U1重叠约45.136秒。输入延迟合计不是campaign墙钟。

主耗时不包含CLI终态后退出等待。模型/工具时间线允许重叠；不得相加当整体墙钟。单trial及跨日历史基线只支持探索性比较。

## 版本计划状态

- 0.8.14.68: complete
- 0.8.14.69: complete

数据：[网页JSON](../exports/web-data.json)、[用户turn CSV](../exports/user-turns.csv)、[模型步骤CSV](../exports/model-steps.csv)、[工具步骤CSV](../exports/tool-steps.csv)。