How it works
The Alt Text Generator audits every image on a page, or in a batch of uploads, and writes a fresh alt-text suggestion for each one using an AI vision model.
URL mode: paste any page URL. We fetch the page through our proxy pool, read the HTML, and find every <img> tag. For each image we capture the context around it: the page title, the H1, and the nearest paragraph or figcaption in the same <figure>. We audit the existing alt for problems (missing, empty, generic, duplicate, too long, keyword-stuffed). Then we send the image and its context to the vision model.
Upload mode: drag and drop or pick image files (JPEG, PNG, WebP, or GIF, up to 10 MiB each). There is no page fetch and no existing alt to audit. The vision model writes alt text from the image alone. Best for new images you have not published yet.
What the vision model returns per image. An alt text of 125 characters or less. A decorative-image classification, so we can recommend an empty alt for pure visual filler. A transcription of any visible text in the image. And a confidence rating of high, medium, or low.
Focus keyword (optional): when you give one, the vision model uses it when it fits the image. The prompt tells the model not to force a keyword that does not match the image. We will not stuff keywords for you.
Language: 12 languages are supported. The alt text comes back in the language you choose, whatever language the page is in.
Plan caps. On the Free plan you can audit up to 5 images per submission and up to 10 images per day. The daily cap exists because each vision call is a paid AI request billed to our account.
What the per-image card shows
- Thumbnail
- A preview of the image being audited. In URL mode it is fetched from the page. In Upload mode it is built in your browser, so the file never leaves your machine until you submit.
- Existing alt
- The current
altattribute on that<img>tag (URL mode only). It is color-coded by quality: green for good, amber for too long or generic, red for missing, empty on a content image, or keyword-stuffed. - Quality badges
- The problems found on the existing alt: missing, empty, generic, duplicate, too-long, keyword-stuffed, or good. Each badge sums up one problem.
- Generated alt (editable)
- The vision model's suggestion. Edit it in place. The character count and the WCAG pass or fail badge update as you type. Aim for under 125 characters.
- Confidence
- The vision model's own rating of its suggestion: high, medium, or low. Low-confidence results deserve a closer look before publishing. The model is flagging an image that is hard to interpret.
- Decorative flag
- When the model classes an image as purely decorative (dividers, watermarks), we recommend
alt=""instead of a description. A "Use empty alt" button appears. - Text in image (OCR)
- Any visible text the model found in the image, written out. Useful for infographics, labelled charts, and accessibility audits of text-heavy images.
- Context (URL mode)
- The text we captured around the image: page title, H1, and the nearest paragraph or figcaption. This is what the vision model saw as context, and it explains why a particular suggestion came back.
Frequently asked questions
What is alt text and why does every image need it?
Alt text, short for "alternative text", is a written description of an image. It goes in the alt attribute of the <img> tag. It serves three audiences. Screen readers read it aloud to blind and low-vision users, so for them the alt text is the image. Search engines use it to understand the image for image search and as a signal about the page. Browsers show it when the image fails to load. The Web Content Accessibility Guidelines (WCAG) require alt text on every meaningful image. A missing alt is one of the most common accessibility failures on the web.
How do I write good alt text?
- Describe what the image shows, in context. "A golden retriever puppy sitting on grass" beats "a dog".
- Keep it under 125 characters. Screen readers may cut off longer text, and search engines may give it less weight.
- Do not start with "image of" or "picture of". Screen readers already say "graphic" or "image" before reading the alt.
- Use keywords only when they describe the image. "Running shoes for flat feet on a beach" is fine. "Best cheap running shoes for flat feet buy now" is keyword stuffing.
- Use an empty alt (
alt="") for purely decorative images such as dividers, watermarks, and ornaments. Screen readers skip them.
Can AI accurately write alt text?
Modern vision models describe image content well. They reliably identify objects, scenes, simple activities, and visible text. They are weaker in three areas. Brand recognition, such as knowing this exact logo is your brand. Fine shades of emotion, such as "delighted" against "amused". And meaning that depends on context, such as "this is the same product as the previous photo". This tool writes a draft alt text per image and shows a confidence rating. Always review and refine before publishing, above all for product pages, journalism, and technical diagrams.
What are decorative images and why is the AI suggesting empty alt for some?
A decorative image adds visual polish but no information: a divider, a background pattern, a watermark, or an icon next to text that already says the same thing. WCAG requires an empty alt (alt="") on these so screen readers skip them. Otherwise a user who cannot see the page hears "image of decorative divider". Our vision audit classifies each image and recommends an empty alt for the ones that look decorative. You can override it before copying.
What do the quality flags on existing alt text mean?
- Missing: there is no
altattribute at all. A WCAG violation. - Empty:
alt="". Fine for a decorative image, a violation on a content image. - Generic: a filename or a vague word such as "image", "photo", or "DSCF1234.jpg". The alt exists but tells screen readers nothing.
- Duplicate: the same alt text on several images on the page, a sign of copy and paste.
- Too long: over 125 characters. Screen readers may cut it off.
- Keyword-stuffed: the same keyword repeated far more than natural.
- Good: passes every check.
Why is the focus keyword option there?
If you give a focus keyword, such as running shoes, the vision model uses it when it fits the image. Upload a photo of running shoes with that keyword and the alt will likely read "Running shoes on a wooden floor" rather than "Athletic shoes on a wooden floor". If the keyword does not fit the image, the model ignores it rather than force it in, which would be keyword stuffing. Leave the field blank and the alt is purely descriptive.
What's the difference between URL mode and Upload mode?
- URL mode: paste any page URL. We fetch the page through our proxy pool and find every
<img>tag. For each image we capture the nearby context (the nearest paragraph, heading, and<figcaption>), audit its existing alt, and write a fresh one. Best for auditing whole live pages. - Upload mode: drag and drop or pick image files. There is no page fetch and no existing alt to flag. The AI writes alt text from the image alone. Best for new images you have not published yet, or images behind a login.
Does this work for images with text in them?
Yes. The vision audit reads any visible text in the image and shows it on the card under "Text in image". That helps with infographics, labelled charts, and meme-style images. WCAG also requires that any important information shown as text inside an image is available as alt text or as visible page text. The transcription makes that check quick.