Download Latest Version llama-b9738-bin-ubuntu-openvino-2026.2-x64.tar.gz (14.1 MB)
Email in envelope

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

Home / b9736
Name Modified Size InfoDownloads / Week
Parent folder
llama-b9736-xcframework.zip < 18 hours ago 206.3 MB
llama-b9736-ui.tar.gz < 18 hours ago 2.7 MB
llama-b9736-bin-win-vulkan-x64.zip < 18 hours ago 38.9 MB
llama-b9736-bin-win-sycl-x64.zip < 18 hours ago 114.3 MB
llama-b9736-bin-win-openvino-2026.2-x64.zip < 18 hours ago 10.5 MB
llama-b9736-bin-win-opencl-adreno-arm64.zip < 18 hours ago 11.7 MB
llama-b9736-bin-win-hip-radeon-x64.zip < 18 hours ago 321.4 MB
llama-b9736-bin-win-cuda-13.3-x64.zip < 18 hours ago 159.5 MB
llama-b9736-bin-win-cuda-12.4-x64.zip < 18 hours ago 261.6 MB
llama-b9736-bin-win-cpu-x64.zip < 18 hours ago 17.3 MB
llama-b9736-bin-win-cpu-arm64.zip < 18 hours ago 11.2 MB
llama-b9736-bin-ubuntu-x64.tar.gz < 18 hours ago 15.6 MB
llama-b9736-bin-ubuntu-vulkan-x64.tar.gz < 18 hours ago 38.5 MB
llama-b9736-bin-ubuntu-vulkan-arm64.tar.gz < 18 hours ago 31.9 MB
llama-b9736-bin-ubuntu-sycl-fp32-x64.tar.gz < 18 hours ago 47.5 MB
llama-b9736-bin-ubuntu-sycl-fp16-x64.tar.gz < 18 hours ago 47.7 MB
llama-b9736-bin-ubuntu-s390x.tar.gz < 18 hours ago 14.7 MB
llama-b9736-bin-ubuntu-rocm-7.2-x64.tar.gz < 18 hours ago 131.3 MB
llama-b9736-bin-ubuntu-openvino-2026.2-x64.tar.gz < 18 hours ago 14.1 MB
llama-b9736-bin-ubuntu-arm64.tar.gz < 18 hours ago 12.6 MB
llama-b9736-bin-macos-x64.tar.gz < 18 hours ago 11.2 MB
llama-b9736-bin-macos-arm64.tar.gz < 18 hours ago 10.9 MB
llama-b9736-bin-android-arm64.tar.gz < 18 hours ago 76.8 MB
cudart-llama-bin-win-cuda-13.3-x64.zip < 18 hours ago 391.0 MB
cudart-llama-bin-win-cuda-12.4-x64.zip < 18 hours ago 391.4 MB
b9736 source code.tar.gz < 19 hours ago 35.0 MB
b9736 source code.zip < 19 hours ago 36.5 MB
README.md < 19 hours ago 4.1 kB
Totals: 28 Items   2.5 GB 0
model : glm-dsa load DSA indexer tensors as optional (#24770) GLM-5.2 ships the DSA "lightning indexer" on only a subset of layers (the "full" layers; others omit it), but the GLM_DSA loader created the five indexer tensors on every layer as required, so loading any GLM-5.2 GGUF failed with e.g. `missing tensor 'blk.3.indexer.k_norm.weight'`. GLM_DSA's graph is llama_model_deepseek2::graph (plain MLA) and does not use the indexer tensors (indexer runtime not yet implemented), so they are loaded-but-unused. Marking them TENSOR_NOT_REQUIRED lets layers without an indexer load as nullptr and the model runs as full MLA attention. DeepSeek-V3.2 (uniform indexer on all layers) is unaffected.

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 2026-06-20