OllamaFrom the builder
Improved performance and model support with GGUF
Ollama 0.30 improves performance and GGUF model compatibility through llama.cpp, complementing its MLX engine on Apple silicon so more models run on more hardware. It is up to 20% faster on NVIDIA hardware, measured with Gemma 4 26B in Q4_K_M quantization on an RTX 5090. Vulkan is now on by default, giving AMD and Intel devices GPU acceleration without vendor-specific libraries.
Vulkan is now enabled by default, extending Ollama’s GPU acceleration to a wider range of hardware, including AMD and Intel devices.
Saved . A shortened excerpt from the official blog; the original publisher retains authorship.