Models & Labs
llama.cpp b10970 Release Expands Platform Support
llama.cpp Releaseshigh confidence
Why it matters
- →Expands platform compatibility, making it easier to deploy AI models across diverse systems.
- →Enhances performance on specific hardware by using fp32 accumulators.
- →Strengthens llama.cpp's position as a versatile inference runtime.
The b10970 release of llama.cpp introduces fp32 accumulation for fattn-mma on CDNA devices, enhancing performance on specific hardware. This update expands support across multiple platforms, including macOS, Linux, Windows, and openEuler, with a focus on CUDA and ROCm libraries. While the release doesn't introduce new models, it strengthens llama.cpp's role as a flexible inference runtime. Developers can now benefit from improved compatibility and performance across a broader range of systems.
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 b10749 Release Expands Platform Support
llama.cpp Releases · September 2, 2026 · Same story
- 2llama.cpp b10933 Release Expands Platform Support
llama.cpp Releases · September 13, 2026 · Same story
More from llama.cpp Releases
Coding Toolscoding
llama.cpp b12022 adds ROCm 10 and Snapdragon support
This release quietly cements llama.cpp as the universal inference runtime by finally bringing ROCm 10.0 to Linux and Windows alongside CUDA 13.4, effectively closing the hardware gap for AMD users who were previously left behind. The inclusion of native Snapdragon support on Linux marks a significant step toward running local LLMs on ARM-based mobile chips without emulation overhead. While KleidiAI on Apple Silicon is disabled in this build, the broader expansion to diverse accelerators ensures that developers aren't locked into NVIDIA ecosystems anymore.
llama.cpp Releases
Coding Toolscoding
llama.cpp b12023 fixes shared prefix bug
This release addresses a specific inference artifact where decision prefixes were incorrectly bleeding into output embeddings. By capping shared prefixes at the first output position, llama.cpp ensures that option tokens remain distinct for parent and child tasks. This is a targeted fix for multi-step reasoning or voting workflows rather than a performance boost. It stabilizes complex generation patterns without changing the underlying model architecture.
