Follow the latest facts and developments on one topic.
The latest release of llama.cpp, version b12132, includes the removal of u8path() in favor of generic_string(), along with updated binaries for multiple platforms including macOS, Linux, Windows, Android, and openEuler, supporting various backends like CUDA, Vulkan, ROCm, OpenVINO, and SYCL.
The latest release of llama.cpp (b12135) introduces support for MEAN operations in WebGPU, enhancing performance and functionality across multiple platforms including macOS, Linux, Windows, Android, and iOS. The release includes precompiled binaries for various architectures and backends such as CUDA, Vulkan, ROCm, OpenVINO, SYCL, and zDNN, with specific builds for Apple Silicon, Snapdragon devices, and openEuler systems. Some features remain disabled or are under development.
The update addresses an issue in llama.cpp where the number of GPU layers was incorrectly calculated by not accounting for skipped NextN layers, leading to improper layer distribution between CPU and GPU. The fix ensures that skipped NextN layers are included in the n_gpu_layers count and properly assigned during model loading, resolving a bug where layer 0 remained on the CPU despite offloading all other layers.
A recent update to llama.cpp addresses a critical issue by rejecting non-object session updates in the server, preventing malformed realtime session data from causing uncaught exceptions and server termination. The release includes binaries for multiple platforms including macOS, Linux, Windows, Android, and iOS, with support for various backends like CUDA, Vulkan, ROCm, OpenVINO, and SYCL.
The latest update to llama.cpp (v1.2140) adds support for the `truncate=false` parameter in YaRN, enabling more accurate frequency interpolation for models like gpt-oss and certain PLaMo-3 variants by preserving non-integer interval boundaries. The change includes updates to GGML operations and CPU backend handling, with CUDA backend support, new tests, and refactoring of RoPE helpers into internal headers. Default behavior remains `truncate=true` for backward compatibility.
The latest update to llama.cpp introduces reference-counted checkpoints instead of deep-copies, reducing memory overhead and improving efficiency. This change adapts the slot save/restore checkpoint mechanism to shared_ptr checkpoint lists and common_state_data buffers, with contributions from multiple developers and AI assistants.
The llama.cpp v1.21.70 release includes enhancements to the moe-cache system, introducing an LRFU/MQ-inspired hybrid cache with improved token handling, key computation, and decay calculations. Changes include adjusting the half-life parameter, increasing the number of classes to 8, refining score computation based on age, removing redundant branches, and improving documentation around unused slots and sorting. The release also features updated build binaries for multiple platforms including macOS, Linux, Windows, Android, and openEuler, with support for CUDA, Vulkan, ROCm, OpenVINO, SYCL, and other backends.
The llama.cpp release b12174 introduces the ability to generate RC files directly without requiring a win-app.rc.in file, alongside a comprehensive set of precompiled binaries for macOS, Linux, Windows, Android, and openEuler across various architectures and backends including CUDA, Vulkan, ROCm, OpenVINO, SYCL, and Snapdragon-specific hardware. The release also includes a standalone UI package.
arXiv:2611.06977v1 Announce Type: new Abstract: Reproducible LLM inference requires that a request produces the same output regardless of the other requests in the batch. Existing solutions pay for this in different ways: batch-invariant kernels constrain every operator, and isolating prefill with fixed-shape decode ({our CoRun-style baseline}) gives up shared prefill compute and pads decode. We present BlockLock, which keeps shared batched prefill and pins only the numerical path of the protected request: it fixes the geometry that the request sees, with {two modes: BlockLock-Rows uses fixed-width rows and retains batched decode of all requests; BlockLock-Pack uses terminal-anchored packing with an isolated dummy segment, then extracts the protected request's KV state for target-only decode. Under a stated per-deployment contract, both modes preserve the target's output bitwise when its companions change.} Across five {dense} models (Qwen3-8B/14B, Llama-3.1-8B-Instruct, and Gemma-3-4B/12B-IT) and three single-GPU deployments (NVIDIA A6000, A100, and H100), we evaluate our approach on both tinyBenchmarks and BFCL{; we also include initial tensor-parallel experiments on two NVIDIA A6000 GPUs. Both modes preserve target sequences and Top20-logit bits in the reported tests.} At equal delivered work, {BlockLock-Rows} reduces latency by 53.5--55.0\% and improves throughput by 115.2--122.2\% relative to vLLM Batch-Invariant; compared with batch-invariant kernels, it needs no custom kernel implementations and constrains only the protected requests.
Shanghai’s Fourier Intelligence built a custom humanoid for Meta’s internal research in months. The company already has a rehab business, modest humanoid shipments, and a disclosed pre-money valuation of about 8 billion yuan. A marquee partner tests its engineering—it does not yet prove the equity story.
The llama.cpp release b12116 introduces support for the drama-3 text-to-speech model, includes cleanup improvements, and ensures the drama3 helper stops at the codebook tower limit. This update provides binaries for multiple platforms including macOS, Linux, Windows, Android, and iOS, with support for various backends such as CUDA, Vulkan, ROCm, OpenVINO, and SYCL. Some builds remain disabled or require additional setup.
A fix was implemented in llama.cpp to correct the calculation of on-device sequence state copy sizes for quantized key-value (KV) caches, ensuring accurate copying of data blocks and preventing crashes or incorrect state restoration. The update adjusts the element count computation using proper block size handling and removes redundant code elements.
The update enables f32 and q5_k multi-column GEMV by default for Adreno X2 on X2E devices, improving performance for specific computational tasks in the llama.cpp framework.
Meta Platforms Inc. has acquired artificial intelligence coding startup Conductor in a deal first reported today by Axios. Conductor confirmed it’s joining Meta in a post on its website. Terms were not disclosed. Conductor’s software lets a developer run several coding agents at once, each in an isolated copy of the codebase, and then review […] The post Meta acquires AI coding startup Conductor, whose app runs several coding agents at once appeared first on SiliconANGLE.
The llama.cpp release b12098 includes accurate support for Qwen3-VL video prompt format and provides precompiled binaries for multiple platforms including macOS, Linux, Windows, Android, and openEuler, with various backend options such as CUDA, Vulkan, ROCm, OpenVINO, SYCL, and Snapdragon-specific optimizations. The release also features a standalone UI package and detailed setup guides for specific hardware configurations.
The llama.cpp release b12109 is now available, offering pre-compiled binaries for macOS, Linux, Windows, Android, and openEuler across various architectures and backends including CPU, CUDA, Vulkan, ROCm, OpenVINO, SYCL, and specialized hardware like Snapdragon. The release includes support for Apple Silicon and Intel macOS, multiple Linux distributions, and detailed setup guides for specific platforms.
The llama.cpp project has been updated to version 0.9.0, with new releases available for multiple platforms including macOS, Linux, Windows, Android, and iOS. The update includes support for various backends such as CUDA, Vulkan, ROCm, OpenVINO, SYCL, and Snapdragon-specific hardware. Some builds are marked as disabled or experimental, and detailed setup guides are provided for specific configurations.
kv-cache : never return 0 kv_rows for sequences without cells (#31570) A cache that shares the cells of another one (f.ex. the gemma4 assistant) can build a graph before its sequences have any cells, f.ex. at the server warmup. get_n_kv_rows() then returned 0, which means that kv_rows are not used, so the graph fell back to the dense KQ mask and the scheduler re-planned for it. Later graphs with larger masks then aborted with GGML_SCHED_DEBUG_REALLOC=1. **Website:** - **Attestations:** - **macOS/iOS:** - [macOS Apple Silicon (arm64)](https://github.com/ggml-org/llama.cpp/releases/download/b12065/llama-b12065-bin-macos-arm64.tar.gz) - macOS Apple Silicon (arm64, KleidiAI enabled) [DISABLED](https://github.com/ggml-org/llama.cpp/pull/23780) - [macOS Intel (x64)](https://github.com/ggml-org/l
Meta has introduced the AI fashion app Style Sketch, which can digitize users’ existing clothing and use AI to generate personalized outfit suggestions. The app is now available on iOS and Android in Indonesia, Singapore, and Thailand, with no global rollout plans announced yet. It supports integration with Instagram, automatically building a virtual wardrobe from photos, and also allows users to manually add items by taking pictures, recording short videos, or uploading images from their photo albums. Additionally, it combines users’ wardrobes with their personal style to recommend coordinated outfits.
Meta stated that, due to strong opposition from publishers regarding AI training terms, it may further postpone the implementation of its new commercial terms—originally scheduled for October 30, 2026, and later extended to November 20. The company emphasized that it will only proceed with policy adjustments after completing comprehensive consultations with industry organizations and leaders.
Why it matters：This news originates from an authoritative tech report published by The Verge on October 28, 2026, which includes Meta’s official statement (Andy Stone’s announcement on X), clear timelines (initially October 30, now tentatively November 20), specific points of contention (authorization for AI model training), and confirmed policy revisions (removal of “training”-related language). The information chain is complete and timely.
