Docker model runner adds Vulkan GPU support (www.docker.com)

Docker has added Vulkan GPU support to Docker Model Runner, its one-command tool for downloading and running local LLMs. Previously limited to CPU, NVIDIA (CUDA), and Apple Silicon (Metal), Model Runner can now leverage Vulkan to accelerate inference on a much broader set of hardware — including AMD, Intel, and integrated GPUs — thanks to integration with the llama.cpp engine. The tool auto-detects compatible Vulkan drivers and will transparently fall back to CPU if none are found, so users don’t need to change configuration to take advantage of GPU acceleration. This is significant because Vulkan is a cross-platform, vendor-agnostic compute and graphics API, so adding support democratizes local model inference beyond NVIDIA- and Apple-centric ecosystems. Practically, it means faster, more energy-efficient LLM runs on many consumer laptops and desktops without needing specialized hardware or complex setup; running a model is as simple as `docker model run ai/gemma3`, which pulls the model and uses your Vulkan-capable GPU if available. Docker Model Runner is open source, inviting community contributions to expand hardware support and features — a meaningful step toward more accessible, performant local AI development.