清华开源 PilotDeck 2.0:智能路由省 83%,跑分超 Claude Code
量子位 · wechat · 2026-10-17
Tsinghua THUNLP, ModelBest and OpenBMB released PilotDeck 2.0, an open-source (AGPL-3.0) agent OS tackling memory management, task execution, and cost.
Key features: WorkSpace-based project isolation, fully white-box memory (inspectable, editable, with Dream-mode background consolidation and rollback), Always-on background task execution, native MCP support across Web/CLI/IM, and Office file processing with checkpoint resume.
Its "strong-main, weak-sub" smart routing scores 70.6 at $3.15 versus single-model Sonnet 4.6's 69.1 at $18.36 — an 83% cost cut. On Workbuddy Bench it scores 79.1 with Qwen3.6-35B-A3B, beating Claude Code's 77.3.
所属事件:清华系团队开源 PilotDeck 2.0,换模型不丢记忆(4 条相关)→
「编程与Agent」频道最新
- 实测 12 个 AI 分类器后,作者直言「分类器新热是cosplay」 — KlausCodes · 2026-10-23
- 用 ESP32 小屏给 Claude Code 做 status 卡片,Go 本地检测状态 — tonylook · 2026-10-23
- Agent 调试新玩法:把 agent 指向 AgentsView CLI 让它自己查 trace 找病因 — dbreunig · 2026-10-23
- YC 系 Mireye 推出土地数据 API+MCP,覆盖 1.6 亿地块 — ycombinator · 2026-10-23
- 实测 12 套决策系统:分类器换版的「LARP 式炒作」陷阱 — KlausCodes · 2026-10-23
- 「给 agent 用的 Product Hunt」上线:按真实调用排出的工具榜 — claudeai · 2026-10-23