Upgrade/Additions to properly run Local Models

A user seeking to run Qwen3.8-Flash on DeepSeek Harness with a 50K context window on a Ryzen 7900X, Radeon 7900XTX, and 96GB RAM asks for inexpensive upgrades and setup advice, noting that others with similar specs have succeeded.

I had Codex help with trying to set up Qwen3.8-Flash on DeepSeek Harness on my main gaming PC. It was giving me some issues with memory and low context windows. What would be an inexpensive upgrade I could do to make this more feasible and would not cause issues on my PC when not running the local model? I currently can run Qwen3.8-27B-Q4 with a 50K context window. Is there a specific way I need to set up flash to run on my current specs? I know others with similar specs have got it working. My current PC specs are as follows CPU: Ryzen 7900X GPU: Radeon 7900XTX RAM: 96GB 2x 48GB G-Skill Trident Z5 6400MHZ usually 70-74GB of RAM is free at any time Windows 11 is there a way to compress its memory usage? Thanks