I checked the documentation of Vercel and here are some advices (founded here and here ):

Latency in Vercel Data Cache: Vercel’s Data Cache is designed to store responses from fetch requests, but each region where your serverless or edge function runs has its independent cache. The latency difference you’re noticing might be due to the time taken to fetch data from the cache layer or the regional edge infrastructure setup. This can include the routing through Points of Presence (PoPs) and communication with the nearest edge region for compute tasks.

Setting Regions for Functions: By default, Vercel's serverless functions run in the iad1 (Washington, D.C., USA) region, which could explain the slower response times if you're in Europe. You can specify a region closer to your data or user base using the regions configuration.

Using Edge Middleware and Network: The Vercel Edge Network leverages PoPs to route user requests to the nearest compute-enabled edge region. While these PoPs minimize the initial latency for requests, the actual compute or cache retrieval may still depend on the selected edge region. If the database or backend server is far from the chosen region, additional latency might occur​.

Cache Configuration: Set a revalidate interval to ensure cached data remains fresh and minimizes origin fetches. Ideally, place your edge region close to your backend or database to avoid cascading latencies

Testing and Optimization: Before production, thoroughly test configurations using Vercel's vercel dev CLI tool to mimic the deployment environment and verify region-specific performance. Consider whether Edge or Serverless runtimes better suit your use case.