Download Latest Version llama-b10329-bin-win-openvino-2026.2.1-x64.zip (80.6 MB)
Email in envelope

Get an email when there's a new version of llama.cpp

Home / b10326
Name Modified Size InfoDownloads / Week
Parent folder
llama-b10326-xcframework.zip < 21 hours ago 277.2 MB
llama-b10326-ui.tar.gz < 21 hours ago 3.1 MB
llama-b10326-bin-win-vulkan-x64.zip < 21 hours ago 34.1 MB
llama-b10326-bin-win-sycl-x64.zip < 21 hours ago 119.5 MB
llama-b10326-bin-win-openvino-2026.2.1-x64.zip < 21 hours ago 80.6 MB
llama-b10326-bin-win-opencl-adreno-arm64.zip < 21 hours ago 12.9 MB
llama-b10326-bin-win-hip-radeon-x64.zip < 21 hours ago 324.6 MB
llama-b10326-bin-win-cuda-13.3-x64.zip < 21 hours ago 146.5 MB
llama-b10326-bin-win-cuda-12.4-x64.zip < 21 hours ago 250.5 MB
llama-b10326-bin-win-cpu-x64.zip < 21 hours ago 18.4 MB
llama-b10326-bin-win-cpu-arm64.zip < 21 hours ago 12.2 MB
llama-b10326-bin-ubuntu-x64.tar.gz < 21 hours ago 16.5 MB
llama-b10326-bin-ubuntu-vulkan-x64.tar.gz < 21 hours ago 32.5 MB
llama-b10326-bin-ubuntu-vulkan-arm64.tar.gz < 21 hours ago 26.5 MB
llama-b10326-bin-ubuntu-sycl-fp32-x64.tar.gz < 21 hours ago 53.0 MB
llama-b10326-bin-ubuntu-sycl-fp16-x64.tar.gz < 21 hours ago 53.2 MB
llama-b10326-bin-ubuntu-s390x.tar.gz < 21 hours ago 15.4 MB
llama-b10326-bin-ubuntu-rocm-7.2-x64.tar.gz < 21 hours ago 130.1 MB
llama-b10326-bin-ubuntu-openvino-2026.2.1-x64.tar.gz < 21 hours ago 101.9 MB
llama-b10326-bin-ubuntu-arm64.tar.gz < 21 hours ago 13.4 MB
llama-b10326-bin-macos-x64.tar.gz < 21 hours ago 11.3 MB
llama-b10326-bin-macos-arm64.tar.gz < 21 hours ago 11.0 MB
llama-b10326-bin-android-arm64.tar.gz < 21 hours ago 76.8 MB
cudart-llama-bin-win-cuda-13.3-x64.zip < 21 hours ago 391.0 MB
cudart-llama-bin-win-cuda-12.4-x64.zip < 21 hours ago 391.4 MB
b10326 source code.tar.gz < 22 hours ago 36.7 MB
b10326 source code.zip < 22 hours ago 38.4 MB
README.md < 22 hours ago 3.9 kB
Totals: 28 Items   2.7 GB 0
tts: account for the vocoder pass in the timings line (#26733) get_output runs the waveform work the pipeline defers to it, from a single trailing window to a full pass depending on the model. Measuring it keeps the reported total and the audio to process ratio honest.

Website:

macOS/iOS:

Linux:

Android:

Windows:

openEuler:

  • DISABLED
  • openEuler x86 (310p)
  • openEuler x86 (910b, ACL Graph)
  • openEuler aarch64 (310p)
  • openEuler aarch64 (910b, ACL Graph)

UI:

Source: README.md, updated 2026-08-07