Image Alt Text Generator

Generate alt text for your images with AI, in English or Spanish. Several at once, editable, with CSV export. Your images are never uploaded.

./alt-text-generator
Alt text language

Checking which AI engine your browser can use…

Images are analyzed inside your browser; they are never uploaded to any server.

What this tool does

You add one or more images and, for each one, an AI model writes a short sentence describing it, ready for the alt attribute of the <img> tag. You choose whether the text comes out in English or Spanish, fix it if needed, copy it, or download all of them together as a CSV with two columns: the file name and its alt text.

Everything happens inside your browser: the images are never uploaded to any server.

How to use it

  1. Choose the alt text language: English or Spanish.
  2. Drag the images in, or click to choose them. You can add more later.
  3. Press Generate alt text. If the model needs downloading, the button tells you its size first.
  4. Review each text: you can edit it, regenerate it or copy it. The counter warns you past 125 characters.
  5. Download the CSV if you are loading the texts into your CMS, your store or a spreadsheet.

Which engine it uses, and how much it downloads

The tool picks between two engines on its own:

  • Chrome’s built-in AI (Gemini Nano), if your browser already has it downloaded and ready. This is the fast path: nothing to download. Today it only exists in desktop Chrome on powerful machines, so most visitors get the other engine.
  • LFM2.5-VL-450M, a vision-language model by Liquid AI that understands the image and writes in English, Spanish and seven other languages. It is downloaded once and kept in the browser cache. It weighs about 510 MB when it runs on the processor, or about 550 MB if your device can use the graphics card (WebGPU), which is also much faster.

It is a big download: on mobile data, better wait for wifi. On a modest computer each image can take several seconds to describe; with a graphics card, much less.

Privacy: the file never leaves your device

The best-known alt text generators upload your image to a server and ask a cloud model for the description. Not here: the model comes to your browser, not the other way round. That matters when the images belong to a client, show an unreleased product or show people. Once the model is downloaded, the tool works even offline.

How to write good alt text (and where the AI helps)

The model describes what it sees. What it does not know is why the image is on your page, and that changes what the right alt text is:

  • Be brief and specific. “Woman reviewing invoices on a laptop” says more than “Image of a person working”.
  • Do not start with “image of” or “photo of”. The screen reader already announces it is an image; the tool strips those openings.
  • Add the context the AI lacks. If the photo shows your product, use its name; if it shows an employee, their name and role.
  • If the image contains important text, include it in the alt. The model is not a good reader of small text: use the image to text extractor for that.
  • Decorative images get alt="", empty. Do not describe them: it only gets in the way of screen reader users.

For SEO, well-written alt text helps the image show up in Google Images and gives the page context; stuffing keywords into it does not help and hurts accessibility.

Limitations

  • It can get people, brands, places and text inside the image wrong. Always review before publishing.
  • Screenshots, charts and diagrams get generic descriptions: write those alt texts by hand.
  • Formats: whatever your browser can open (JPG, PNG, WebP, GIF, AVIF). Large images are scaled down before analysis, so there is no need to resize them.

Frequently asked questions

What is the alt text of an image?

It is the text in the alt attribute of the img tag. Screen readers read it to people who cannot see the image, the browser shows it if the image fails to load, and search engines use it to understand what the image is about. A good description is short and says what matters about the image in that context.

Are my images uploaded to a server?

No. The AI model is downloaded to your browser and analyzes the images right there. The only thing that travels over the network is the model download, once; your images never leave your device.

How big is the download, and why?

About 510 MB (around 550 MB if your device uses the graphics card). It is a vision-language model that has to understand the image and write in your language, and that does not fit in less. The tool tells you the size before downloading anything, and it stays in the browser cache for next time. If your Chrome already has its built-in AI (Gemini Nano) ready, that is used instead and nothing is downloaded.

How long should alt text be?

There is no official limit, but the usual advice is to stay under about 125 characters: some screen readers cut off or tire the listener with long texts. That is why the counter warns you when you go over. If an image needs a long explanation (a chart, a diagram), keep the alt short and put the explanation in the page text.

Can I use the text as is?

As a starting point, yes; as the final version, review it. The AI describes what it sees, but it does not know why the image is on your page, and it can get names of people, brands, places or text inside the image wrong. Decorative images should have an empty alt, not a description.