RTX 3090 (24 GB VRAM) GPU and/or its dock. You can connect it to your laptop using high-speed USB Type-C (USB4 or Thunderbolt); or you can put it inside your PC.
I have added a fan to the backside to protect it from accidental touching and to keep it cool.
It is great for running local AI models such as Qwen-3.8-27B, Gemma-4-31B, financial models, diffusion models, etc. It uses Cuda, which is essential for compatibility with most AI models.
24 gb is plenty to fit most models.
