| Entity | Relation | Why |
|---|---|---|
| 2604.14228 (Claude Code) | 优化对象参照 | Claude Code 是 production multi-step agent pipeline,AgentOpt 的 model combination selection 直接适用于优化 Claude Code 中 sub-agent delegation 的 model assignment |
| 2511.00739 (CPU-Centric) | 互补视角 | CPU-Centric 刻画 tool execution 占 E2E 延迟 88%,AgentOpt 优化 model selection 占据的 cost budget——两者在不同资源维度(compute vs cost)优化同一 pipeline |
| 2603.18897 (PASTE) | 延迟互补 | PASTE 通过 speculative tool execution 减少 agent 端到端延迟 48.5%,AgentOpt 通过 model selection 减少 cost 62–76%——延迟优化 vs 成本优化 |
AgentOpt 发现 model quality is not context-free——Claude Opus 4.6 作为 standalone 最强但作为 planner 最差 [2604.06296]。这直接挑战了 Claude Code 的设计假设(使用单一 frontier model for all roles)[2604.14228]。PASTE 和 CPU-Centric 优化 pipeline 执行效率,AgentOpt 优化 pipeline 组合配置——三者在同一 agent system 的不同优化维度上操作。
| 维度 | AgentOpt 视角 | Claude Code 设计 | ||
|---|---|---|---|---|
| Model assignment | 组合搜索($ | M | ^N$ 空间) | 单一模型 (Claude Sonnet for all roles) |
| 优化粒度 | Pipeline-level end-to-end metrics | Per-interaction quality | ||
| 核心发现 | Best standalone model ≠ best per-role model | "1.6% 决策逻辑" = model choice is fixed | ||
| Sub-agent delegation | Different model per role is optimal | Same-model delegation (Claude Code spawns Claude Code) |
核心矛盾: Claude Code 的 "minimal scaffolding" 哲学认为 single frontier model 足以处理所有 roles [2604.14228],AgentOpt 实证证伪了这一假设——Opus 作为 HotpotQA planner 仅 31.71% (bypasses solver tools),而 Ministral 3 8B + Opus 达到 74.27% [2604.06296]。
| 维度 | AgentOpt | CPU-Centric |
|---|---|---|
| 优化目标 | Cost/accuracy Pareto | Latency/throughput |
| 瓶颈识别 | Model selection (13–32× cost gap) | Tool execution (88% of E2E latency) |
| 方法 | Bandit-based combination search | COMB micro-batch overlap + MAS hybrid scheduling |
| 作用时间 | 离线(搜索阶段)→ 固定部署 | 在线(运行时调度优化) |
互补性: AgentOpt 选择成本最优组合(离线优化),CPU-Centric 加速选定组合的运行时执行(在线优化)[2511.00739]。两者可顺序应用:先 AgentOpt 选组合,再 COMB/MAS 优化该组合的 execution scheduling。
| 维度 | AgentOpt | PASTE |
|---|---|---|
| 优化目标 | Cost (62–76% savings) | Latency (48.5% reduction) |
| 优化维度 | Model assignment (which model for which role) | Execution timing (overlap tool exec with LLM thinking) |
| 约束 | Fixed pipeline structure | Fixed model + tool set |
| 方法 | Pure-exploration bandit (UCB-E) | Pattern Tuple + speculative execution |
| 正确性 | End-to-end accuracy preserved | No correctness impact (speculation + promotion) |
互补性: PASTE 假设 fixed model assignment 优化 execution overlap [2603.18897]。AgentOpt 可先确定 cost-optimal combination,然后 PASTE 在该 combination 上进一步优化 latency——形成 cost-first → latency-second 的两阶段优化。
AgentOpt 在 agent optimization stack 中的定位:
Architecture design (Claude Code) → Model assignment optimization (AgentOpt) → Runtime scheduling (PASTE, CPU-Centric) → Infrastructure serving (Autellix, vLLM)
AgentOpt 填充了"架构确定后、部署执行前"的优化空白。其核心发现(model quality is not context-free)对整个 agent 设计社区有范式意义:per-call routing (RouteLLM style) 在 multi-step pipelines 中是 fundamentally wrong abstraction [2604.06296]。
采纳信号: Open-source (GitHub), framework-agnostic (httpx interception 适用于任何 Python agent framework)。但仅验证了 2-role pipelines;real-world agents 可能有 5–10 roles。