Browsing

AI Agents

How AI agents work, what they cost to run and which model fits each step, backed by our own benchmarks and builds.

2 posts

All posts

Every Model Has a Battlefield: Typed-Decision Models vs LLMs in Agent Loops

In a sequential agent loop, a typed-decision model won all 10 Wikipedia races against 7 LLMs: 7.3s of model time vs 17.2s for Gemini Flash-Lite.

"Just add memory" is the reflex fix when an agent forgets everything between sessions, and it's the wrong instinct. There isn&