The Bottom Line: A free and open-source AI image upscaler for Windows, macOS, and Linux that runs entirely on your own GPU via Vulkan acceleration, with no cloud upload, no account, and no subscription for the core desktop app. It uses Real-ESRGAN and related models to reconstruct detail in blurry photos, compressed JPEGs, pixel art, and anime illustrations, supporting up to 16× upscaling. Multiple models are included for different content types, along with Batch Upscayl for folder queues and Double Upscayl for a second processing pass on difficult sources. The app is AGPL-licensed, the GitHub repository has roughly 41k stars, and the current desktop version is v2.11.0. A separate cloud version exists with credit-based paid plans, but the desktop app remains free with no paywall on core upscaling.

Pricing model: Free and open source for the desktop app. The official FAQ states plainly that Upscayl is "100% free and open-source" with "no premium paywalls, sudden subscription models, or hidden processing fees" . A separate cloud offering exists with credit-based plans, but the core desktop upscaler carries no cost.

Platforms: Windows, macOS, Linux. Native installers are provided for Windows 10/11, macOS 12+, and Linux (AppImage, with Flatpak, Snap, DEB, and RPM also available) . No mobile app or browser-based version is documented for the desktop app.

Pros & Cons

Pros: The desktop app is genuinely free and open source (AGPL) with no paywall, watermark, or usage limits on core upscaling. The official FAQ confirms there are no hidden fees for high-resolution rendering . All processing happens locally on your GPU via Vulkan, so images never leave your machine—the features page explicitly states "100% offline after models are installed" with "no uploads and no cloud queue delays" . It supports multiple AI models tuned for different content, including Real-ESRGAN for general photos, anime-specific models for line art, and UltraSharp for sharpening, plus Batch and Double Upscayl for production workflows . Cross-platform availability covers Windows, macOS, and Linux with native installers and package manager support .

Cons: GPU compatibility is limited. The GitHub README warns that "you'll need a Vulkan compatible GPU" and that "many iGPUs (integrated graphics) do not work" . The troubleshooting guide is more blunt: "Upscayl does not work without a GPU. CPU and iGPU-only systems are unsupported" . This means some laptops and lower-end desktops may not run it at all. Results are not magic—the FAQ notes that while it can enhance low-resolution and pixelated images, it "cannot de-blur or do focus adjustment" on out-of-focus or severely blurred photos . The interface and settings are simpler than commercial tools like Topaz, with fewer fine-grained controls for advanced retouching workflows. The cloud version and third-party domains create confusion about what is official, and there is no mobile app.

Who is it for?

Ideal for: Photographers and archivists restoring old or low-resolution photos who want local processing without uploading to cloud services; designers and content creators upscaling screenshots, UI elements, logos, and social media images for print or high-resolution displays; artists and animators upscaling illustration, anime line art, and pixel art with content-specific models; developers and game creators batch-upscaling textures and legacy assets on their own hardware; privacy-conscious users and teams handling confidential visual assets who cannot use cloud upscalers.