Microsoft 论文：会话历史可提升多轮 Agent 的模型路由效果
微软这篇论文给做 Agent 降本的提了个实用思路：路由时把会话历史喂给路由器，多轮任务多过 3.3 个点，还把切换的缓存成本算清楚了。
微软一篇论文研究多轮 Agent 场景下的成本感知模型路由。他们在多轮表格 Agent 上测试零样本路由，在相同卸载率下，读取会话历史的路由器比只看当前请求的多通过 3.3 个百分点的任务。论文还量化了切换成本：从贵模型切到便宜模型，两次调用内就能摊销缓存成本，反向切换则无法回本。长会话中便宜模型的成本优势从 6.8 倍降到 3.2 倍。
Interesting paper from Microsoft on context-guided model routing.
They found a nice trick to achieve cost-aware model routing for agents in a multi-turn setting at the user-turn level.
If you route agent turns between a cheap model and an expensive one, they suggest giving the router the session history.
Microsoft tested zero-shot routing on multi-turn spreadsheet agents.
At the same offloading rate, a router that reads the history passes 3.3 points more tasks than one that sees only the current request.
Prompt caching adds a cost to every switch. Switching from the expensive model to the cheap one recovers its cache cost within two calls. Switching up did not break even in their tests.
The cheap model's cost advantage also shrank from 6.8x to 3.2x on longer sessions.
https://t.co/vuLndvoLWa
Chat with Paper: https://t.co/cNRFiaE7PW
