Models & Labs
llama.cpp b10658 release adds DFlash2 support
llama.cpp Releaseshigh confidence
Why it matters
- →DFlash2 support enhances local convolution and candidate selection, improving model performance.
- →Code optimizations and bug fixes contribute to a more stable and efficient runtime.
- →The update supports a wide range of platforms, increasing accessibility for developers.
The b10658 release of llama.cpp brings DFlash2 support, focusing on local convolution and candidate selector enhancements. Assisted by Claude Opus 5, the update includes optimizations like cost reduction and code refactoring, alongside bug fixes and formatting improvements. These changes aim to enhance performance and stability across multiple platforms, including macOS, Linux, Windows, and openEuler. This release signifies a step forward in making llama.cpp a more efficient and reliable tool for developers.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1Llama.cpp b10105 Release Refines Argument Handling
llama.cpp Releases · July 25, 2026 · Same story
What happened next
- 2llama.cpp v0.4.0 Release
llama.cpp Releases · September 5, 2026 · Same story
More from llama.cpp Releases
Coding Toolscoding
llama.cpp b12336 fixes HunyuanVL crash
This release quietly patches a critical memory overflow in the mtmd module that caused crashes when loading HunyuanVL models. The issue stemmed from F64 position embeddings overflowing an F32 buffer, a subtle bug that broke multimodal inference for users relying on this specific architecture. By enforcing strict F32 requirements, llama.cpp stabilizes support for Tencent's vision-language model without adding new features or changing the core API. It’s a necessary maintenance update that keeps local multimodal workflows from breaking unexpectedly.
llama.cpp Releases
Models & Labsother
llama.cpp b12340 adds CUDA 13 and ROCm 10.0
This release quietly closes the hardware gap for local inference by adding default support for CUDA 13 and ROCm 10.0 across Linux and Windows. Users with newer NVIDIA GPUs or AMD accelerators no longer need to compile from source to get performance, while Apple Silicon builds now ship KleidiAI disabled by default, shifting the focus back to standard ARM optimizations. The inclusion of Snapdragon support for Linux and Android marks a significant step toward on-device AI beyond the x86/ARM desktop ecosystem. It is less about new features and more about ensuring llama.cpp remains the universal runtime for every available accelerator.
