Upgraded my local setup with 2 rtx pros and it's amazing.

A user upgraded their local setup with two RTX Pros, significantly improving performance for AI models. They noted that while Qwen consumes much context with "thinking tokens," Deepseek 1M context allows for four concurrent requests and ample VRAM. The user appreciates the speed of both models, even if they make errors, and is seeking suggestions for new models or installations.