AI tools that run inside your browser
Upscale, cut out, transcribe and speak. Free and unlimited, and your files never leave your device.
Tools that run a model on your device
Each model downloads once, caches on disk, then runs on your GPU. Nothing is sent anywhere.
Background Remover
Erase the background from any image.
Image Upscaler
Sharpen and enlarge small images up to 4x.
Image to Text (OCR)
Pull text out of screenshots and photos.
Subject Cutout
Cut any object out as a transparent PNG.
Audio Transcription
Turn audio or video into timestamped text.
Text to Speech
Turn text into speech and download a WAV.
Voice Cloning
Record a short sample, then speak in that voice.
AI Video Clipper
Cut a long video into short captioned clips.
Text-Based Video Editor
Edit video by deleting words in its transcript.
Background Noise Remover
Remove hum, hiss and room noise from audio.
Redact Text
Hide personal details before pasting into AI.
Filler Word Remover
Cut the ums and uhs from a recording.
Small everyday utilities
No download. Opens instantly, works offline.
Your files never leave the tab
Open the network panel while a tool runs. After the model weights come down once, there is nothing left to see. No upload endpoint exists to send a file to.
Models execute on your graphics card through the browser's own runtime, not on a rented GPU somewhere.
Weights land in IndexedDB and OPFS on the first run, so the second run starts with no download at all.
There is no login and no usage meter, because there is no server cost per run to recover.
Whisper is MIT and Qwen3.5 is Apache 2.0. Each model is named where it is used.
One email when a new tool ships
Roughly once a month. Unsubscribe in one click.