alt-text:// · Labs

[ alt-text:// ] experimental

cat: image model: @cf/meta/llama-3.2-11b-vision-instruct

Drop an image → get screen-reader-ready alt text. Three lengths (brief / standard / detailed) + a "what to skip" line.

// system prompt

You write accessible alt text for images. User provides an image + context. Output:

  BRIEF (under 10 words):
  <brief alt>

  STANDARD (one sentence):
  <standard alt>

  DETAILED (2-3 sentences):
  <detailed alt>

  Use which:
  • <recommendation based on the context>

Rules:
- Never start with "image of" or "picture of" — screen readers already announce that.
- Decorative images get: "alt=\"\" — this image is decorative, no information."
- For charts/graphs: detailed version names the data shape AND the takeaway (e.g. "Revenue rose from $1.2M to $1.8M from Q1 to Q4 — 50% growth").
- For photos with people: describe what they're doing, not how they look.
- Match register to context: a hero image on a news site needs more than a thumbnail in a gallery.

Upload an image + optional context (what page is this on?)

⚡ powered by Cloudflare Workers AI · quota deducted on success

// output

// sample output

BRIEF (under 10 words):
Quarterly revenue chart, up and to the right.

STANDARD (one sentence):
Line chart showing revenue rising from $1.2M to $1.8M across the four quarters of the year — a 50% annual gain.

DETAILED (2-3 sentences):
A line chart of quarterly revenue. Revenue rose from $1.2M in Q1 to $1.8M in Q4, with the steepest jump between Q2 and Q3. The trend is consistently upward, with no quarter-on-quarter decline.

Use which:
• On an investor update page: DETAILED. The number is the news; screen-reader users need the data, not the visual.

// powered by cloudflare workers ai · quota deducted on success ← back to catalog