Strata 0.2.12 is a new release that allows users to run the Qwen3.8-Flash-Next model on gaming PCs equipped with just a 12 GB graphics card, significantly improving performance on budget GPUs like the RTX 3060. The update includes a re-tuned DFlash2 helper model and various fixes for coding agents, enhancing tool-calling capabilities. Benchmark tests have shown up to a threefold increase in prompt-speed on specific hardware, while also providing improvements for AMD and Apple devices. The release consists of 42 files totaling 9.70 GiB, available for download from GitHub, with known issues and limitations transparently outlined

Strata 0.2.12 Targets AMD, Apple, and Budget NVIDIA GPUs

Strata 0.2.12 lets you run Qwen3.8-Flash-Next on a gaming PC with just a 12 GB graphics card. The release delivers its biggest speed gains on budget GPUs, adding up to roughly a threefold prompt-speed jump on RTX 3060-class cards. It also re-tunes its DFlash2 helper model and rolls out a wave of tool-call fixes aimed at coding agents like Claude Code and OpenCode. Download 42 files totaling 9.70 GiB from the official GitHub release to get started.

Strata 0.2.12 Targets AMD, Apple, and Budget NVIDIA GPUs @ Linux Compatible