10-19-2026, 02:53 AM
Did a small comparison this afternoon. I couldn't reproduce the slowdown until I pushed context much higher. The surprise was that memory pressure showed up before throughput collapsed.
routing beats arguing over one perfect model
