WebNN
From U-Net to DiT: Z-Image Turbo Runs in Your Browser
Z-Image Turbo is a 6B-parameter Scalable Single-Stream Diffusion Transformer (S3-DiT) running entirely in the browser via WebGPU. Intel Web Platform Engineering adapted the model through ONNX conversion, INT4/FP16 quantization, and operator fusion — achieving a 7x size reduction and up to 7x inference speedup for real-time, on-device text-to-image generation on AI PC hardware. Read more
ONNX2WebNN - Reducing Web AI Framework Overhead by 99%
WebNN is a privacy-first browser API enabling hardware-accelerated, client-side AI inference. While frameworks offer convenience, they inflate bundle sizes (2-20 MB) and slow startup. Alternatively, using WebNN directly in vanilla JavaScript yields minimal overhead (tens to hundreds of KB). Tools like ONNX2WebNN and browser‑based code generators automate converting ONNX models to optimized WebNN code, supporting layout tuning and quantization. Read more