b9670

ggml-org/llama.cppb9670Jun 16, 2026by github-actions[bot]

AI Summary

A bug fix release addressing edge cases in NVFP4 (NVIDIA Floating Point 4) quantization within the llama-graph implementation. The changes ensure correct order of operations for LoRA and ModelOPT, specifically moving post-GEMM MUL operations.

Key Highlights

  • Fix for NVFP4 edge-cases in llama-graph
  • Correction of post-GEMM MUL order for LoRA and ModelOPT
  • Restriction of build_ffn for NVFP4 to supported combinations

New Features

  • NVFP4 quantization bug fixes

Full Release Notes

<details open>

Fix and restrict NVFP4 edge-cases in llama-graph (#24331)

* Move post-GEMM MUL required for dequant b4 lora and bias add

see https://github.com/ggml-org/llama.cpp/pull/23484 :
1. For lora, I would presume we want fully dequantized values before
   doing the residuals, but this depends on how the LORAs were
generated. Literature tells me LORA happens post-mul but pre-bias add https://github.com/ggml-org/llama.cpp/pull/8332
2. For ModelOPT, bias-add should happen on [fully-dequantized
   values](https://github.com/NVIDIA/Model-Optimizer/blob/b49f9b9e2d747af992d78a3aa7f10efe5a8847e1/modelopt/torch/quantization/backends/nvfp4_gemm.py#L59-L64)

* Restrict build_ffn for NVFP4 to supported combinations

</details>

**macOS/iOS:**
- [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-macos-arm64.tar.gz)
- macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780)
- [macOS Intel (x64)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-macos-x64.tar.gz)
- [iOS XCFramework](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-xcframework.zip)

**Linux:**
- [Ubuntu x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-ubuntu-x64.tar.gz)
- [Ubuntu arm64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-ubuntu-arm64.tar.gz)
- [Ubuntu s390x (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-ubuntu-s390x.tar.gz)
- [Ubuntu x64 (Vulkan)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-ubuntu-vulkan-x64.tar.gz)
- [Ubuntu arm64 (Vulkan)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-ubuntu-vulkan-arm64.tar.gz)
- [Ubuntu x64 (ROCm 7.2)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-ubuntu-rocm-7.2-x64.tar.gz)
- [Ubuntu x64 (OpenVINO)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-ubuntu-openvino-2026.0-x64.tar.gz)
- [Ubuntu x64 (SYCL FP32)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-ubuntu-sycl-fp32-x64.tar.gz)
- [Ubuntu x64 (SYCL FP16)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-ubuntu-sycl-fp16-x64.tar.gz)

**Android:**
- [Android arm64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-android-arm64.tar.gz)

**Windows:**
- [Windows x64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-win-cpu-x64.zip)
- [Windows arm64 (CPU)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-win-cpu-arm64.zip)
- [Windows x64 (CUDA 12)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-win-cuda-12.4-x64.zip) - [CUDA 12.4 DLLs](https://github.com/ggml-org/llama.cpp/releases/download/b9670/cudart-llama-bin-win-cuda-12.4-x64.zip)
- [Windows x64 (CUDA 13)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-win-cuda-13.3-x64.zip) - [CUDA 13.3 DLLs](https://github.com/ggml-org/llama.cpp/releases/download/b9670/cudart-llama-bin-win-cuda-13.3-x64.zip)
- [Windows x64 (Vulkan)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-win-vulkan-x64.zip)
- [Windows x64 (SYCL)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-win-sycl-x64.zip)
- [Windows x64 (HIP)](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-bin-win-hip-radeon-x64.zip)

**openEuler:**
- [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23705)
- openEuler x86 (310p)
- openEuler x86 (910b, ACL Graph)
- openEuler aarch64 (310p)
- openEuler aarch64 (910b, ACL Graph)

**UI:**
- [UI](https://github.com/ggml-org/llama.cpp/releases/download/b9670/llama-b9670-ui.tar.gz)