AI Briefing#webgpu
The latest AI and developer news about #webgpu, with the original source and a short summary.
The latest AI and developer news about #webgpu, with the original source and a short summary.
The new backend executes torch.export .pte files directly in Chrome, Safari, and other WebGPU-enabled browsers without format conversion.
The new backend allows a single .pte file to run on Chrome, Safari, and other WebGPU browsers without conversion to other formats.
Fine-tuned on 2B tokens of synthetic data, the model predicts screen deltas from keypresses, outperforming prompted baselines on shell and file readback tasks.
The 2B parameter model achieves up to 500 tokens per second locally in the browser on an M4 Max using custom WebGPU kernels.
Framefields is an open-source, Apache-2.0 licensed npm package that enables code-first video creation by combining Photoshop-inspired compositing, After Effects-style motion graphics, and Blender-like 3D rendering into a single WebGPU-based framework designed for AI coding agents.
Gitframes offers a code-first video rendering solution that bypasses headless browsers to achieve 60–120+ FPS and low memory usage, integrating directly with coding agents like Claude Code and Cursor.
LiquidAI has released LFM2.5-Encoder-350M and LFM2.5-Encoder-230M, with the 350M variant described as a multilingual bidirectional encoder supporting 15 languages and on-device inference.
The collection supports more than 200 common ML operations running entirely locally in the browser via WebGPU.
The open-source project replicates the 71-block Swin/ViT architecture of DLSS-NR build 310.8.0 using FP8 tensor cores, achieving byte-for-byte intermediate matches with the original.
The project introduces webbrain-vl-2-450M, a specialized vision model trained on browser screenshots that runs locally via WebGPU.
Three.js r186 integrates 3D Gaussian Splatting as a first-class engine feature, enabling realistic 3D rendering via WebGPU.
WebLLM, which runs LLMs in the browser via WebGPU without a server, has been released with OpenAI API compatibility.
Hugging Face has released 207 optimized WebGPU kernels and a loader library, improving browser-based AI inference performance by an average of 2.57x.
Hugging Face has released 207 optimized WebGPU kernels and a loader library to enhance in-browser AI inference performance.
Superwhisper's 600M parameter LLM, S1-mini, runs locally in the browser via WebGPU.
jax-js has released a browser-based ML library that leverages WebGPU and WebAssembly.
The ggml-webgpu implementation in llama.cpp significantly improved inference speed through matmul optimization for k-quants models.
PrismML has unveiled Bonsai Image 4B, a 1-bit/ternary-based model that can run locally in the browser via WebGPU.
webml-kit has been released, making it easy to run ML models in the browser using WebGPU and WASM.
Outlines the structure for attaching local AI to Manifest V3 Chrome extensions using Transformers.js.
A Chrome extension that runs Gemma 4 on-device inside the browser.
An AVBD-based rigid-body/soft-body physics prototype was implemented with WebGPU.
Transformers.js v4 introduces a WebGPU backend, enabling high-performance AI model execution in both browser and server environments.