Models & Labs
llama.cpp adds Prism Bonsai 2 27B support
llama.cpp Releaseshigh confidence
Why it matters
- →Prism Bonsai 2 27B is now runnable locally without custom conversion scripts.
- →Native Hadamard support ensures efficient inference on diverse GPU backends.
- →Expands the library of specialized models available for consumer-grade hardware.
llama.cpp has released support for the Prism Bonsai 2 27B model, integrating native runtime capabilities for its specific architecture. The update addresses technical requirements such as Hadamard tensor handling and weight folding, enabling efficient inference across major platforms including CUDA 12/13, ROCm 10.0, and Apple Silicon. This addition allows users to deploy the 27B parameter model directly via standard llama.cpp binaries without additional preprocessing steps. The release underscores the growing availability of specialized, high-efficiency models optimized for local hardware constraints.
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
- 1PrismML compresses LLMs to run locally with minimal loss
TechCrunch AI · September 17, 2026 · Same story
- 2llama.cpp b11081 release with CUDA 13 and ROCm 10 support
llama.cpp Releases · September 22, 2026 · Same story
- 3llama.cpp b11378 adds ROCm 10 and CUDA 13
llama.cpp Releases · October 4, 2026 · Same story
Follow this story
Open the full story →llama.cpp b11222 release with ROCm 10 and CUDA 13
2 developments
- Sep 28 · llama.cpp Releases
llama.cpp b11222 release with ROCm 10 and CUDA 13 - Oct 12 · llama.cpp Releases
llama.cpp adds Prism Bonsai 2 27B support (This article)↳ llama.cpp adds native support for Prism Bonsai 2 27B with runtime optimizations for Hadamard transformations
More from llama.cpp Releases
Coding Toolscoding
llama.cpp b11555 adds ROCm 10 and Snapdragon support
This release quietly cements llama.cpp as the universal inference runtime by finally bringing first-class ROCm 10.0 support to both Linux and Windows. AMD GPU users no longer need workarounds, effectively closing a long-standing parity gap with NVIDIA's CUDA ecosystem. The inclusion of Snapdragon AI stack binaries for Linux marks a significant expansion into mobile-edge hardware, while the addition of CUDA 13 builds future-proofs the tool against upcoming driver requirements. By standardizing these complex backend integrations into single downloadable binaries, the project removes friction for anyone deploying local models on non-NVIDIA silicon.
llama.cpp Releases
Coding Toolscoding
llama.cpp RPC logging overhaul
This release refactors llama.cpp’s RPC backend logging, turning a binary debug flag into a granular verbosity system. Developers can now trace tensor transfers and graph cache evictions with precision, which is critical for debugging distributed inference setups. The cleanup also unifies log handling across client and server sides, reducing noise and fixing format specifier bugs. It’s a maintenance win that makes diagnosing remote inference failures significantly easier.
llama.cpp Releases
Coding Toolscoding
llama.cpp b11558 fixes OpenCL and adds ROCm 10
This release quietly patches critical stability issues for OpenCL inference on Qualcomm hardware, ensuring q4_k dense bin kernels no longer crash when image limits are exceeded. It also brings ROCm 10.0 support to both Linux and Windows, keeping the project aligned with AMD's latest driver stack. While KleidiAI is temporarily disabled on macOS builds, the core value here is fixing fragmentation for non-NVIDIA users who rely on Vulkan or OpenCL for local LLMs.
llama.cpp Releases
More in Models & Labs
© The AI Daily BriefModels & Labsmodels
Anthropic Launches Claude Dashboards and Motion
Anthropic has released new enterprise features including Claude Dashboards for usage monitoring and Motion for workflow automation.
The AI Daily Brief
© The Rundown AIModels & Labsagents
Microsoft's Copilot pivot: Product over models
Microsoft is quietly admitting that raw model intelligence isn't enough to drive adoption. By stripping low-value entry points from Windows and shifting to a multi-model routing strategy, they are prioritizing product form factors over benchmark chasing. This signals a broader industry correction where utility beats novelty, forcing labs to prove their models fit into actual workflows rather than just winning leaderboards.
The Rundown AI
© Lev SelectorModels & Labsmodels
Mistral Large 4 and Claude Haiku 5.5 Released
Mistral releases Large 4 'Le Chonk' while Anthropic launches Claude Haiku 5.5, continuing the trend of cheaper, faster frontier models.
Lev Selector