Image Generation

What it is

Claude doesn’t generate photos or illustrations the way a dedicated image tool does. There’s no “create an image of…” pipeline that returns a bitmap.

What Claude can do is build diagrams, charts, and interactive visuals inline in the conversation, using HTML and SVG. These are live — you can click, adjust, and iterate in follow-up turns. That feature is in beta on Claude web and desktop. See Visual and interactive content for the full story.

Claude can also analyse images you upload — see Add files or photos for upload mechanics.

Where to find it

On desktop

Anywhere in chat. Ask for a diagram, chart, or visualisation and Claude may build one inline.

On the web

Same — any chat on claude.ai.

How to use it

Ask for the kind of visual you want: “draw this as a flowchart”, “chart this data”, “show me a side-by-side comparison”.
Claude decides whether a visual helps and renders it inline.
Iterate in conversation — “make it blue”, “add a third column”, “flip the axis”.
To keep a visual, use Copy as image, Download (SVG/HTML), or Save as artifact.

When you need a real image

If you need a photo or an illustrated scene, use a dedicated image-generation tool. Claude can help you write the prompt for it, but it won’t produce the pixels.

Good to know

Custom visuals are beta. Quality varies. Opus handles complex visualisations best.
Not on mobile. Custom visuals render on web and desktop only.
Can’t save by default. Custom visuals are ephemeral — they live in the chat. Save them as an artifact if you want them to stick around.
Uploading images works fine. Claude can read, describe, and reason about images you upload, even though it can’t paint new ones.

What’s changed recently

2026-04-18 — Page created from official help-center article.