Strata released version 0.1.44 of its open-source inference engine on October 13, with a hotfix following just five hours later. This update allows the software to run the 125-billion-parameter Qwen3.8-Flash-Next model on a single consumer GPU, while also introducing native support for Apple Silicon Macs, alongside enhancements for NVIDIA GPUs. The update includes first-class support for Unsloth quantized formats and the NVFP4 packs specifically for NVIDIA cards, improving performance significantly compared to previous versions. Additionally, the hotfix addressed critical bugs that affected system compatibility and performance, ensuring that answers remain consistent with earlier builds

Strata 0.1.44.1 Adds Mac Support, Unsloth Quants, and NVFP4

Open-source inference engine Strata released version 0.1.44 on October 13, followed by a hotfix just five hours later. The MIT-licensed tool, built by a developer known only as "Niko," runs Qwen3.8-Flash-Next, a 125-billion-parameter Alibaba model, on a single consumer GPU with 12 GB of VRAM. The new versions add native Apple Silicon support for M1 through M5 Macs, first-class Unsloth quantized formats, and NVIDIA-only NVFP4 packs, all while keeping answers mostly byte-identical to earlier builds.

Strata 0.1.44.1 Adds Mac Support, Unsloth Quants, and NVFP4 @ Linux Compatible