Models & Labs
llama.cpp b9850 Release Enhances Qwen3 Model
llama.cpp Releaseshigh confidence
Why it matters
- →Enhancements in Qwen3 model could lead to better performance and reliability.
- →Broad platform support increases the framework's versatility.
- →Fixes and additions strengthen the existing model infrastructure.
The b9850 release of llama.cpp focuses on refining the Qwen3 model with fixes in input assignment and the addition of a tensor for attention normalization. This update aims to enhance the model's performance and reliability. It also expands platform support, covering macOS, Linux, Windows, and openEuler, although some features like KleidiAI on Apple Silicon remain disabled. This release is a step towards improving the framework's adaptability and efficiency across various computing environments.
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 b9495 release enhances platform support
llama.cpp Releases · June 4, 2026 · Same story
- 2llama.cpp b9653 Release Expands Platform Support
llama.cpp Releases · June 16, 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.
