b8987
108K-star LLM inference engine gets new release with updated dependencies
The llama.cpp open-source project, with 108K stars and 17.6K forks on GitHub, released version b8987, a maintenance update that upgrades the embedded HTTP library (cpp-httplib) to version 0.43.2 for improved security and stability. The release, signed by a Hugging Face contributor, provides extensive prebuilt binary support across all major platforms: macOS (Apple Silicon with optional KleidiAI acceleration, Intel x64, iOS XCFramework), Linux (x64 and arm64 CPU, plus Vulkan, ROCm 7.2, OpenVINO, SYCL FP32/FP16), Windows (x64/arm64 CPU, CUDA 12/13, Vulkan, SYCL, HIP), and Android (arm64 CPU). Additionally, openEuler Linux is supported with specific hardware (310p, 910b) using ACL Graph.
This release continues llama.cpp's tradition of enabling efficient local LLM inference on consumer hardware, from laptops to high-end GPUs. The vendor update addresses potential vulnerabilities in the networking layer used for server mode, while the broad binary distribution eliminates the need for manual compilation. The lack of major new features suggests this is a stability release, but the comprehensive testing across 20+ build configurations underscores the project's commitment to reliability across diverse hardware ecosystems.
- Updated cpp-httplib to 0.43.2 for enhanced security and reliability
- Prebuilt binaries for macOS, Linux, Windows, Android, and openEuler with multiple GPU backends (Vulkan, ROCm, CUDA, SYCL, HIP, OpenVINO)
- Supports Apple Silicon with KleidiAI acceleration and iOS XCFramework
Why It Matters
Maintains llama.cpp as the go-to local LLM runner across all platforms with critical library updates.