
Native AI image generation for Windows 11 ARM64, optimized for Snapdragon hardware using ONNX Runtime and the Phoenix Engine.
HK NPU STUDIO is an independent source-available proprietary desktop application for local generative AI on Windows 11 ARM64 PCs. It is built specifically around Snapdragon X hardware and aims to make local AI image generation approachable without forcing users to understand model package layouts, QNN components, ONNX files, or installation internals. The Phoenix Engine coordinates model discovery, validation, installation, activation and inference while keeping supported generation workflows local on the PC. The application supports Windows 11 ARM64 and targets Snapdragon X Plus / X Elite systems, with integrated Qualcomm QNN / Hexagon NPU and ONNX Runtime. Supported model workflows include Stable Diffusion 1.5, Stable Diffusion 2.1, and Stable Diffusion 3.5 Medium, while SDXL and additional models are experimental. It includes a Model Manager for guided installation, validation and activation. Phoenix Boost improves or expands prompts before image generation. Deterministic mode works locally without an additional language model. Phoenix Image Lab includes AI Photo Restoration, which restores historical, damaged, and black-and-white photos, preserves available detail, removes conservative dust and speck defects with Dust Cleanup V4, and produces a 4× upscaled result. All FiDeSR neural restoration stages use Qualcomm QNN/HTP on the Snapdragon Hexagon NPU, and the required Photo Restore models are included in the RC3 installer.
Users use HK NPU STUDIO to generate AI images locally with direct control over models, prompts, presets, resolution, generation parameters, and advanced options.
Users use HK NPU STUDIO to restore historical, damaged, and black-and-white photos with 4× upscaling on the Snapdragon NPU.
Users use Phoenix Boost to improve or expand prompts before image generation.
Users use ControlNet Canny to guide image generation using structural edge information.
Users use the Model Manager to install, manage, and review supported AI models inside the application.
Users use the gallery to browse and manage locally generated images in one central gallery.
Users use image comparison to compare generated images side by side with synchronized navigation, zoom, and panning.
Users use the guided first run to move from first launch to first image.
Lynote combines AI-draft rewriting and AI-text detection with note-taking, transcription, video summaries, and study tools.
LMSYS Chatbot Arena is a crowdsourced open platform for LLM evals. Collected over 1,000,000 human pairwise comparisons to rank LLMs with the Bradley-Terry model and display the model ratings in Elo-scale.
Zhipu Qingyan is a GLM-based AI assistant whose official description emphasizes understanding goals, breaking down tasks, and using tools.
字节跳动AI助手