Google open-sourced ML Drift under Apache 2.0 as the GPU engine behind LiteRT and as a standalone library. It targets OpenCL, Metal, WebGPU and OpenGL ES with tensor virtualization, runtime shader generation, custom operators and 5D tensor support.
Why it matters: A shared engine across mobile, browser and desktop GPUs could reduce the backend-specific work needed to ship local generative and vision workloads. Its programmable shader surface also creates a new performance and correctness boundary for custom operators.
Signed prereleases through b11046 add or repair CUDA-only builds, Vulkan IQ3_S kernels, graph-buffer reservation failures, WebGPU strides, model-saver sliding-window-attention patterns, allocation checks and OpenCL Q6_K layout compatibility, then extend Hexagon IM2COL and ROLL operations and add a guarded OpenCL flash-attention kernel.
Why it matters: The release train now spans allocation and model-conversion correctness plus new Hexagon and OpenCL execution paths. Each change is hardware- and graph-specific, so a later build can fix one backend while creating a new qualification surface elsewhere.
Signed llama.cpp prereleases b10919 through b10936 add multi-device Qualcomm Hexagon support, WebGPU and OpenCL correctness fixes, server download-state repairs, an internal JSON-schema representation, structured JSON logging and better complex-type parsing for Qwen3-Coder tool calls. Builds b10932-b10936 also repair cached clang precompiled headers and Jinja integer-property parsing.
Why it matters: The sequence now spans both accelerator correctness and the schema-to-grammar path used for structured tool calls. Local-model operators need to qualify output, tool arguments and observability together instead of treating a successful model load as…
Hugging Face released an Apache-2.0 preview of 207 WebGPU kernels as individually versioned Hub repositories, a JavaScript loader and Fleet, an in-browser correctness and performance test suite. Each kernel packages its contract, WGSL implementation, correctness cases and benchmark cases.
Why it matters: Browser inference can now depend on inspectable, versioned low-level operations instead of opaque or unversioned shaders. That creates a reproducible path for testing local AI performance and correctness across the device, browser and driver combinations…
Liquid AI released a 3.1-billion-parameter vision-language model for screen understanding, grounding, multi-image work, and function calling. Weights and day-one llama.cpp, MLX, vLLM, SGLang, ONNX, GGUF, and WebGPU paths target phones, laptops, single GPUs, and local browser use.
Why it matters: A compact visual model with broad runtimes can keep GUI, OCR, and multimodal tool workflows closer to private data and end devices.