flâneur

Computer use is 45x More Expensive Than Structured APIs

reflex.dev · 1,428 words · saved by 1 readers

We benchmarked computer use against auto-generated API endpoints on the same admin panel. 53 steps and 551k tokens vs 8 calls and 12k tokens.

Blog Open Source We ran a benchmark comparing two ways of letting an AI agent operate the same admin panel, with the goal of putting a price tag on vision agents (browser-use, computer-use). Here is what we measured, what we had to change to make the vision agent work at all, and what changes when generating an API surface stops being a separate engineering project. Why vision agents? Vision agents are the default for letting AI agents operate web apps that don't expose APIs. The alternative, writing an MCP or REST surface per app, is its own engineering project across the 20+ internal tools m

saved by

related reading