Strata is free, open-source software that runs a 125-billion-parameter AI model (Qwen3.8-Flash-Next) on consumer gaming PCs with a single NVIDIA card (12-24 GB VRAM) and 64 GB RAM, generating text at 60-95 tokens per second. It supports multiple GPU configurations, various model compressions, and specialized variants like a coding-focused version, with experimental AMD Radeon support.