AI Alt Text Generator
Drop in your images and a real AI captioning model writes the alt text for you. It runs entirely on your own machine, so nothing is uploaded. Edit any description, then copy it or export them all.
🔒 100% private — the model runs locally, your images never leave your deviceDrop images here
or click to choose files · PNG, JPG, WebP, GIF · as many as you like
How does AI run in your browser?
This page loads a real image-captioning neural network — a Vision Transformer that looks at the picture, paired with a GPT-2 text decoder that turns what it sees into a sentence. It is fetched once with Transformers.js and then runs on your own CPU through WebAssembly, inside a background worker so the page never freezes. There is no API key and no server doing the work. After the first download the model is cached, so it keeps working even with your internet off.
Why on-device matters for alt text
Most alt-text tools send your images to someone else's server. For private photos, client screenshots, or anything under NDA, that is a non-starter. Here the picture is read, downscaled, and described all on your machine, then thrown away when you close the tab. You get the speed of a local tool and the privacy of never uploading anything.
Good alt text, faster
Accessible images need short, accurate descriptions, and writing them by hand for a whole gallery is tedious. The AI gives you a solid first draft for every image in seconds. You skim, fix the few it gets wrong, and move on. Keep it under roughly 125 characters, describe what matters in the image rather than starting with "image of", and leave purely decorative images with empty alt text.
Are my images uploaded anywhere?
No. The model runs locally in your browser through WebAssembly. Your images are downscaled and described on your device and never sent to a server.
How big is the download and is it slow?
The model is a few hundred megabytes, downloaded once and cached by your browser. The first image takes a moment while the model warms up; after that each description takes a few seconds on a normal laptop.
Can I edit what it writes?
Yes. Every description is a normal editable text box. The AI gives you a first draft, you refine it, then copy it or export everything as CSV or JSON.
How accurate is it?
It is a compact open model, so it is good at the main subject of a photo but can miss small details or text in the image. Treat it as a fast first draft, not a final answer — always read it before you ship it.
Does it work offline?
Yes, once the model has downloaded the first time. Everything after that runs without a network connection.
I build tools like this every day.
Senior full-stack engineer, available for senior or contract work, fully remote. See the rest of the lab or get in touch.