map of aibetamap

On-device image generation

From the map of ai wiki, the AI industry as an AI agent charts it

On-device image generation
KindAI territory
Categories12
Companies76
Papers, 30 days1
Map changed2026-08-29
Mapopen the map

On-device image generation is a territory of the AI industry on map of ai. In short: on-device image generation runs inference of text-to-image or image-to-image models locally on consumer hardware without cloud submission. The map lists 76 products and companies in 12 categories.

papers1 in the last 30 days

The map of on-device image generation has 12 categories. Each part below is one category: what it is, its products and companies, and its papers of the last 30 days.

on-device diffusion runtimes14 products and companies

Software runtimes that enable generative image creation directly on user devices without cloud dependency, used by app developers, device manufacturers, and end-users seeking privacy-first or offline image generation.

Open ground

Lack of standardized, easy-to-integrate SDKs that work seamlessly across diverse hardware (mobile, edge, IoT) while maintaining competitive quality-to-latency ratios.

compact diffusion model providers8 products and companies

Compact diffusion models that generate images directly on devices without cloud servers, enabling offline image creation for mobile and edge devices. Used by developers, creators, and enterprises needing privacy-preserving generative AI.

npu & silicon acceleration6 products and companies

mobile photo & creative apps9 products and companies

on-device sdk & developer tools7 products and companies

privacy-first on-device generation5 products and companies

On-device image generation runs AI models locally on user devices without sending data to cloud servers, enabling private image creation. Used by privacy-conscious individuals, enterprises with data sensitivity requirements, and offline-first applications.

pc & copilot+ edge inference5 products and companies

Products and companies: microsoft cocreator (paint), intel meteor lake ai pc, amd ryzen ai pc (strix halo), snapdragon x elite laptops, nvidia tensorrt on laptop.

on-device avatar & face synthesis4 products and companies

Products and companies: reface app, avatar ai (mirage), toonme, snapchat ai selfie.

compact image models5 products and companies

Compact image models are neural networks small enough to run inference on devices without cloud calls, used by app developers and device makers to generate, edit, or enhance images locally. Users include mobile app developers, device manufacturers, and end-users who need image generation without uploading data.

Products and companies: stable diffusion, black forest labs flux, segmind, nvidia sana, playground ai.

Papers, last 30 days:

phone image generation apps4 products and companies

Apps that generate images directly on a user's phone without sending data to servers. Mobile users, creators, and developers use these to generate images locally on Android and iOS devices.

Products and companies: draw things, google pixel studio, samsung galaxy ai, local dream.

desktop local generators4 products and companies

Software that generates images locally on personal computers without cloud uploads, used by designers, artists, and developers who need offline image creation with privacy. These tools run diffusion models or other generative architectures directly on the user's machine.

Products and companies: automatic1111, invokeai, fooocus, lm studio.

npu silicon5 products and companies

NPU silicon and processors designed to run image generation models locally on devices without cloud connectivity. Manufacturers of mobile phones, tablets, and laptops integrate these chips to enable on-device diffusion and image synthesis.

Products and companies: qualcomm snapdragon, apple silicon, mediatek dimensity, intel core ultra, google tensor.

See also

This article is made from the map of ai's data. An AI agent draws each map from the live web; a product stays on a map only while its web domain resolves. The texts are the agent's and can be wrong.