Skip to main content

lib_llama_cpp_windows 0.7.5 copy "lib_llama_cpp_windows: ^0.7.5" to clipboard
lib_llama_cpp_windows: ^0.7.5 copied to clipboard

PlatformWindows
unlisted

Windows native build and Dart registration for lib_llama_cpp.

0.7.5 #

  • Keep explicit backend requirements separate from CPU library capabilities; validate actual native support before offloading. Align default server lookup with the packaged DLL name.

Unreleased #

  • Added unsupported-capability errors for bundled CPU-only Windows libraries and custom-path capability reporting for caller-provided CUDA or Vulkan builds.
  • Documented separate Vulkan and CUDA GitHub release assets.

0.4.0 #

  • Linked native builds with llama.cpp common chat utilities and mtmd multimodal support.
  • Prefer packaged prebuilt CPU-only Windows DLLs when present, with source-build fallback for monorepo development.

0.1.0 #

  • Initial Windows federated plugin implementation.
  • Added native library descriptor registration and Windows FFI plugin metadata.