Download Latest Version llama-b8683-bin-ubuntu-openvino-2026.0-x64.tar.gz (77.0 MB)
Email in envelope

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

Home / b8680
Name Modified Size InfoDownloads / Week
Parent folder
llama-b8680-xcframework.zip < 17 hours ago 175.9 MB
0
llama-b8680-bin-win-vulkan-x64.zip < 17 hours ago 56.6 MB
0
llama-b8680-bin-win-sycl-x64.zip < 17 hours ago 135.5 MB
0
llama-b8680-bin-win-opencl-adreno-arm64.zip < 17 hours ago 33.3 MB
0
llama-b8680-bin-win-hip-radeon-x64.zip < 17 hours ago 361.3 MB
0
llama-b8680-bin-win-cuda-13.1-x64.zip < 17 hours ago 168.5 MB
0
llama-b8680-bin-win-cuda-12.4-x64.zip < 17 hours ago 252.3 MB
0
llama-b8680-bin-win-cpu-x64.zip < 17 hours ago 39.4 MB
0
llama-b8680-bin-win-cpu-arm64.zip < 17 hours ago 32.2 MB
0
llama-b8680-bin-ubuntu-x64.tar.gz < 17 hours ago 31.7 MB
0
llama-b8680-bin-ubuntu-vulkan-x64.tar.gz < 17 hours ago 48.8 MB
0
llama-b8680-bin-ubuntu-vulkan-arm64.tar.gz < 17 hours ago 41.0 MB
0
llama-b8680-bin-ubuntu-s390x.tar.gz < 17 hours ago 35.0 MB
0
llama-b8680-bin-ubuntu-rocm-7.2-x64.tar.gz < 17 hours ago 168.9 MB
0
llama-b8680-bin-ubuntu-openvino-2026.0-x64.tar.gz < 17 hours ago 77.0 MB
0
llama-b8680-bin-ubuntu-arm64.tar.gz < 17 hours ago 27.8 MB
0
llama-b8680-bin-macos-x64.tar.gz < 17 hours ago 104.2 MB
0
llama-b8680-bin-macos-arm64.tar.gz < 17 hours ago 40.3 MB
0
llama-b8680-bin-910b-openEuler-x86-aclgraph.tar.gz < 17 hours ago 72.5 MB
0
llama-b8680-bin-910b-openEuler-aarch64-aclgraph.tar.gz < 17 hours ago 64.9 MB
0
llama-b8680-bin-310p-openEuler-x86.tar.gz < 17 hours ago 72.5 MB
0
llama-b8680-bin-310p-openEuler-aarch64.tar.gz < 17 hours ago 64.9 MB
0
cudart-llama-bin-win-cuda-13.1-x64.zip < 17 hours ago 402.6 MB
0
cudart-llama-bin-win-cuda-12.4-x64.zip < 17 hours ago 391.4 MB
0
b8680 source code.tar.gz < 18 hours ago 29.7 MB
0
b8680 source code.zip < 18 hours ago 30.9 MB
0
README.md < 18 hours ago 3.5 kB
0
Totals: 27 Items   3.0 GB 0
[CUDA ] Write an optimized flash_attn_stream_k_fixup kernel (#21159) * Write an optimized flash_attn_stream_k_fixup kernel Write a specialized and more optimized kernel for cases where nblocks_stream_k is multiple of ntiles_dst. Make nblocks_stream_k to multiple of ntiles_dst if nblocks_stream_k > 2 * ntiles_dst * Use the new kernel only for nblocks_stream_k_raw > 4 * ntiles_dst to make sure we have enough concurrency on GPUs * Address review comments * Address review comments * Revert variable names to original

macOS/iOS:

Linux:

Windows:

openEuler:

Source: README.md, updated 2026-04-06