ProductsSelected

Nace AI 发布 Drex Router:按步骤为智能体工作流动态分配模型

Why we picked it

搞智能体的朋友可以看看这个:Nace AI 的 Drex Router 会按每一步任务挑模型,200 个智能体建 SaaS 从 $9.48 砍到 $1.31,代码迁移比纯用 Opus 5.5 便宜 38%,正确率只差 1 个点。预算敏感的 agent 开发者值得测一下。

Nace AI 推出 Drex Router,一个面向智能体工作流的逐步模型路由器,根据每个步骤的难度把任务分配给合适的模型。

Body

在 Claude Code 修 bug 场景中,Drex Router 用时 1:19、花费 $0.57,而只用 Opus 需 2:42 和 $0.

91。在代码迁移任务中,API 成本比 Opus 5.

5 降低 38%,文件移植正确率为 92.3%,仅略低于纯 Opus 的 93.

3%。在 200 个智能体协作的 SaaS 构建测试中,成本从 $9.

48 降到 $1.31,降幅约 86%。

随着低价闭源模型和开源前沿模型同时增多,路由层正成为成本与质量的关键瓶颈,Drex Router 也将进入竞争日趋激烈的模型路由市场。

Original · Aran Komatsuzaki (论文推介)

Nace AI presents Drex Router, a per-step model router for agent workflows:

- Claude Code bug fix: 1:19 vs 2:42, $0.57 vs $0.91 vs Opus-only - Code migration: -38% API cost vs Opus 5.5, 92.3% vs 93.3% files ported correctly - 200-agent SaaS build: $9.48 -> $1.31

With strong low-cost closed models and powerful open frontier models becoming increasingly available, routing is becoming a key bottleneck. Drex Router is an important addition to the increasingly competitive router market.

Hope to see more evals on non-coding tasks, which tend to be underserved.

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Products" desk, focusing on US frontier AI labs and Silicon Valley technology leaders. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows and US-China foundation model benchmarks and commercialization comparison. Key entities and technical topics under continuous tracking include Drex Router, Nace AI, Claude Code, Opus.

Source attribution: curated from Aran Komatsuzaki (论文推介) (@arankomatsuzaki).