llama.cpp Releases
684 releases of ggml-org/llama.cpp
- b10330
Optimizes CUDA performance by fusing RMS normalization, multiplication, and RoPE operations into a single kernel launch.
Aug 8, 2026
- b10329
Updates the Server and UI to only display the working directory control when at least one enabled tool actually requires it.
Aug 8, 2026
- b10328
Adds initial support for tool isolation using Docker containers to enhance security and sandboxing.
Aug 8, 2026
- b10327
Fixes a bug in CUDA kernel launches for quantized copy operations where the thread/block count was incorrectly calculated.
Aug 8, 2026
- b10326
Improves the accuracy of Text-to-Speech (TTS) timing reporting by accounting for the vocoder pass in the pipeline.
Aug 7, 2026
- b10322
Optimizes the SYCL backend performance for SSM (State Space Model) convolution operations by coalescing window loads.
Aug 7, 2026
- b10321
Fixes a critical correctness bug in the Metal backend where NORM and RMS_NORM operations produced incorrect results for rows with partial SIMD groups.
Aug 7, 2026
- b10319
Fixes a bug in the multi-modal text module where the `longest_edge` parameter was being ignored when combined with min/max pixel settings.
Aug 7, 2026
- b10318
A general sync release for the ggml library, providing pre-built binaries for a wide range of platforms including macOS, Linux, Windows, Android, and openEuler.
Aug 7, 2026
- b10313
This release adds an LRU (Least Recently Used) scheduler to the server's router component. This allows for better management of model evictions and waiting queues.
Aug 7, 2026
- b10312
This release modifies the server router's eviction policy to prevent evicting busy models. This ensures that models currently in use are not unloaded from memory unnecessarily.
Aug 7, 2026
- b10311
This release fixes a bug in the Multi-Modal Text-to-Speech (MTMD) pipeline for Qwen3-TTS. The issue caused the model to feed the text stream twice during generation, leading to the talker reading the utterance twice.
Aug 7, 2026
- b10310
This release improves ARM64 CPU support by adding fallback definitions for missing HWCAP bits and fixing the detection of the FP16 variant. It also renames has_fp16_va to has_fp16 for clarity.
Aug 7, 2026
- b10308
A bug fix release addressing a crashing issue on Windows MSYS2 UCRT64 environments running GCC 16.1.0.
Aug 7, 2026
- b10307
This release fixes a critical parsing bug in the SYCL backend where NVFP4 quantization format scaling factors were incorrectly being treated as signed values instead of unsigned values.
Aug 7, 2026
- b10306
A performance-focused release for SYCL backends, specifically optimizing the GLU (Gated Linear Unit) operations. It introduces a flat path for contiguous memory and refactors kernel launches for better performance on Arc Pro B70.
Aug 7, 2026
- b10305
This release adds support for specific DSv4 (Data Shape v4) operations in the SYCL backend, including LIGHTNING_INDEXER and various DSV4_HC variants.
Aug 7, 2026
- b10303
A quick follow-up release to fix a SYCL error related to the OP_FLASH_ATTN_EXT operation on the Arc770 GPU architecture. It maintains the same binary distribution as the previous release.
Aug 7, 2026
- b10301
A maintenance release focusing on fixing CUDA compiler warnings regarding unused variables and functions. It includes standard binaries for macOS, Linux, Android, Windows, and openEuler across various architectures and hardware backends.
Aug 7, 2026
- b10299
This release fixes a compilation error in the Metal backend where declaring arrays of matrix types in `threadgroup` memory was not supported. The fix uses a POD array and casting.
Aug 7, 2026