小墨模型评测横向归档(截至 2026-05-29)
打开 HTML ↗
xiaomo-model-eval-cross-model-archive按方案、评测、调研、选型、复盘等用途沉淀长期内容。短回复留结论,完整证据、判断链和执行记录放到这里,方便检索、回看和继续迭代。
总览、选型、方案、调研、复盘和关键验证优先。
xiaomo-model-eval-cross-model-archivehermes-shared-memoryagentmemory-vs-hindsight-memory-selectionaegis-partial-absorption-plangenericagent-absorption-evalalbert-lsk-wechat-notebank61 篇
xiaomo-model-eval-detail-gpt-5-4-mini-xhighxiaomo-model-eval-detail-gpt-5-5-highxiaomo-model-eval-detail-gpt-5-5-xhighxiaomo-model-eval-detail-gpt-5-4-mini-highcpa-codex-reasoning-effort-aliasessharedchat-codex-adapter-cpa-hermesxiaomo-model-eval-detail-yt-claude-4-7xiaomo-model-eval-detail-yt-claude-4-6xiaomo-model-eval-detail-skyclaw-v1xiaomo-model-eval-detail-qwen3-7-maxxiaomo-model-eval-detail-mimo-v2-5-proxiaomo-model-eval-detail-kimi-k2-6xiaomo-model-eval-detail-gpt-5-5xiaomo-model-eval-detail-gpt-5-4xiaomo-model-eval-detail-gpt-5-4-minixiaomo-model-eval-detail-gpt-5-4-2026-03-05xiaomo-model-eval-detail-gpt-5-3-codexxiaomo-model-eval-detail-glm-5-1xiaomo-model-eval-detail-gemini-3-5-flashxiaomo-model-eval-detail-gemini-3-1-pro-previewxiaomo-model-eval-detail-deepseek-v4-proxiaomo-model-eval-detail-deepseek-v4-flashxiaomo-model-eval-detail-claude-sonnet-4-6xiaomo-model-eval-detail-claude-opus-4-8xiaomo-model-eval-detail-claude-opus-4-7xiaomo-model-eval-detail-claude-opus-4-7-thinkingxiaomo-model-eval-detail-ali-qwen-3-6xiaomo-model-eval-detail-ali-kimi-k2-5xiaomo-model-eval-cross-model-archivespice-json-publish-checkscript-publish-checkhtml-anything-display-validation-2026-05-23hermes-capability-bundles-convergencehermes-529-upgrade-assessmentaegis-partial-absorption-planaegis-partial-absorption-executionaegis-absorption-evaluationspice-json-publish-checkspice-json-direct-publish-checkspice-decision-layer-lite-absorption-final-reportspice-decision-card-field-absorb-checkscript-publish-checkhermes-github-pr-issue-trenddecision-layer-lite-e2e-smoketextscript-publish-checkhtmlfeishu-long-reply-auto-remediationalbert-lsk-wechat-notebankscript-publish-checkdecision-trace-second-checkdecision-trace-project-split-checkcpa-vs-sub2api-selectionagentmemory-vs-hindsight-memory-selectionhermes-shared-memoryhermes-project-workspace-layerfeishu-long-reply-guard-testhtml-anything-display-validation-2026-05-23genericagent-vs-hermes-lcm-prompt-compactiongenericagent-absorption-evaldecision-reply-html-archive-sample长期入口规则
新增内容:用 decision_trace_publish.py 生成 HTML/wiki/public,刷新 Hub 后部署 Pages。
排序口径:全部归档按记录日期倒序排列;文件 mtime 只作为同日内辅助,不作为跨日期主排序。
视觉版本:HTML Main Visual v1.0。