AI NEWS JOURNAL

AI, IN PERSPECTIVE

Search AI news

Search headlines, summaries and articles to connect the dots.

2026-11-08

llama.cpp releases05:11

b12191

The latest update to llama.cpp introduces enhanced MoE (Mixture of Experts) cache functionality, enabling computation of cache misses on the CPU with a configurable CPU share. A new tuning script, moe-cache-tune.py, automatically determines optimal cache size and CPU share settings based on system performance. The update also improves scheduling efficiency, removes redundant sync operations, and refactors code for better maintainability, while maintaining compatibility across multiple platforms including macOS, Linux, Windows, and Android.

llama.cpp releases05:28

b12192

The latest release of llama.cpp adds support for the MageVL-Cowork model and adjusts how rope frequencies are handled, now stored at the model level within clip_model instead of being copied into each clip_layer. The loader verifies that rope_freqs is a 32-bit floating-point array with dimensions matching d_head, ensuring compatibility with expected 4:6:6 splits in the computational graph.

llama.cpp releases07:45

b12193

The llama.cpp release b12193 includes a change to move log-mel computation to the cgraph, removes stray files, updates mtmd_audio_track to include source, and adds tests. It provides precompiled binaries for macOS, Linux, Windows, Android, and openEuler across various architectures and backends including CUDA, Vulkan, ROCm, OpenVINO, SYCL, and Snapdragon-specific platforms.

llama.cpp releases09:11

b12195

The llama.cpp release b12195 introduces a change to refuse device buffers that would exceed VRAM, improving memory management. It includes precompiled binaries for multiple platforms including macOS, Linux, Windows, Android, and openEuler, with support for various backends such as CUDA, ROCm, Vulkan, OpenVINO, SYCL, and others. Some builds are marked as disabled or experimental.

llama.cpp releases09:29

b12196

The release includes updates for SYCL backend with exact rounding support in CPY operations for Q4_1, Q5_1, and Q8_0 quantization formats.

llama.cpp releases09:46

b12197

The update introduces support for automatic loading of libhtp without relying on legacy environment variables, enhances path handling for non-ASCII directories and long paths, adds platform-wide support for FASTRPC_MANAGE_DSP_LIBRARY_PATH and GGML_HEXAGON_LIBHTP_DIR, maintains forward slashes in absolute paths on Android and Linux, ensures absolute path precedence with fallback mechanisms, and includes CDSP1,2,3 in the fallback chain for IQ10 compatibility.

llama.cpp releases10:05

b12198

The llama.cpp release b12198 includes a change to downgrade per-request proxy logs to trace level, along with numerous precompiled binaries for macOS, Linux, Windows, Android, and other platforms, supporting various backends like CUDA, Vulkan, ROCm, OpenVINO, and SYCL. Some builds are disabled or experimental, including openEuler and KleidiAI-enabled macOS versions.

llama.cpp releases12:43

b12199

The release addresses the issue of invalid UTF-8 characters in arguments passed to PEG tool calls, improving input handling and robustness.

llama.cpp releases14:39

b12200

Enhancements to OpenCL support in llama.cpp include adding new kernel functions for quantized matrix operations, improving small-n MoE GEMV performance across multiple quantization types, optimizing Adreno GPU kernel floor settings, renaming functions for clarity, and disabling bin FA for small verification batches.

llama.cpp releases14:58

b12201

The llama.cpp project has released version b12201, offering pre-compiled binaries for multiple platforms including macOS, Linux, Windows, Android, and iOS. The release includes support for various backends such as CUDA, Vulkan, ROCm, OpenVINO, SYCL, and Snapdragon hardware. Notably, the MoE cache feature is disabled for the draft model in this release. Additional resources include setup guides for Snapdragon devices and separate downloads for CUDA and OpenVINO libraries.

llama.cpp releases16:11

b12202

The llama.cpp v1.220.2 release introduces enhanced BF16 precision support across multiple components, aligning with reference implementations by moving specific operations like GLM4 family projections, MiMoVL vision matmuls, and Qwen3-TTS speaker encoder conv1d to BF16 where appropriate. Backends such as CUDA, Hexagon, OpenCL, WebGPU, and Vulkan now uniformly handle GGML_PREC_BF16 similarly to F32, while CPU, SpaceMIT, and ET backends extend flash attention support to BF16 with F32 accumulation. Flash attention tests now include BF16 cases, and the KQ multiplication in Llama is updated to request BF16 precision. A TODO remains in clip.cpp for future work.

llama.cpp releases22:20

b12207

A new fast path for d_state=64 has been added to the SSM_SCAN kernel in ggml-cuda, improving generation throughput for Zamba2-7B-Instruct from 11 to 15.1 tok/s—approximately a 35% increase—by eliminating CPU fallbacks for this state size. The change was verified with real-world Zamba2 hyperparameters and maintains correctness, though a separate 28% of scan calls still fall back to CPU due to an undiagnosed issue.

llama.cpp releases23:47

b12208

The ggml library has been updated to make tied selections in the radix Top-K algorithm deterministic on CUDA and Metal backends by replacing atomic compaction loops with a deterministic scheme using simdgroup scans and coalesced memory accesses. This change ensures byte-identical outputs across runs, improves performance by 20-45% on M5 Ultra hardware, and enables deterministic testing for Metal and ROCm. Redundant CUDA kernel resets were removed, and Vulkan/Metal changes were reduced or skipped.

2026-11-07

llama.cpp releases00:07

b12175

A fix was implemented to override the hardware_concurrency value in CI tests, addressing an issue where test-llama-context and test-backend-ops hung due to incorrect thread count estimation on 8-CPU runners. The solution sets default values of 8 for TEST_ARG_HARDWARE_CONCURRENCY and 4 for TEST_ARG_PARALLEL_WORKERS specifically for the cpu-x64-high-perf configuration.

llama.cpp releases00:39

b12176

The update qualifies calls to math functions like isnan and signbit with std:: to ensure compatibility with C++ standard library overloads, addressing issues on NetBSD 11.99.9/x86_64 with gcc 14.4.0. Additional C math function calls are standardized using std:: prefixes (e.g., std::fabs, std::copysign) instead of non-existent std::fabsf variants, improving accuracy by keeping computations in double precision. The change results in a minor behavioral difference with a relative error of at most 1.4e-7, well below test tolerances.

llama.cpp releases01:09

b12177

A fix has been implemented to prevent crashes in the command-line interface when encountering invalid UTF-8 in prompts or files, with assistance from Claude. This update is part of the b12177 release of llama.cpp, which includes various binary distributions for macOS, Linux, Windows, Android, and other platforms.

llama.cpp releases04:43

b12179

The update introduces support for 3D convolution (conv_3d) on Hexagon hardware by generalizing the existing conv_2d implementation. It leverages the HMX tiled path, folds kernel dimensions into a virtual row, and uses a shared solver with an is3d flag. The host computes unfold divisors to avoid device-side division, prevents DMA queue overflow, and improves tile size selection based on VTCM usage. Additional changes include handling arbitrary positive strides, preserving load/shuffle paths for small strides, and fixing compiler warnings.

llama.cpp releases07:18

b12180

The release adds the Kimi-K3.1-Pro text model to llama.cpp, replacing full-attention MLA layers in K3 with DSA (Dynamic Sparse Attention) layers featuring a lightning indexer. The indexer uses an additive mask over the cache and reuses sparse flash attention paths, while preserving existing KDA layers and MoE components. The model's architecture includes four indexer tensors per DSA layer, hybrid memory for caching, and maintains quantization integrity by excluding indexer parameters from quant overrides. Verified against HF checkpoint with identical PPL beyond context length, and supports multiple platforms including macOS, Linux, Windows, Android, and iOS.

llama.cpp releases11:30

b12182

The latest release of llama.cpp (v1.2182) includes a change to raise the compute view limit for buffers with few tensors in the meta backend, along with reverting the deepseek41 test model back to a 6-layer configuration. The release provides precompiled binaries for multiple platforms including macOS, Linux, Windows, Android, and iOS, supporting various backends such as CUDA, Vulkan, ROCm, OpenVINO, SYCL, and Snapdragon-specific hardware. Some openEuler builds remain disabled.

llama.cpp releases17:17

b12183

The release includes the addition of arange functionality for OpenCL, enhancing support for array operations in the OpenCL backend. This update is part of a broader set of improvements and compatibility fixes across multiple platforms, including macOS, Linux, Windows, Android, and various GPU and NPU backends.