Qwen 3.5 27B Runs Full 256K Context Under 20GB VRAM on RTX PRO 6000

Qwen 3.5 27B in NVFP4 w/ full context taking less than 20GB VRAM You can basically run like 5 agents w/ full context on a single RTX PRO 6000 like this, and they'd be so fast Tell me I didn't tell you this was gonna happen