Short answer: no, Claude cannot generate photos or AI art the way ChatGPT, Gemini or Midjourney can. Anthropic’s own help page says so plainly.
But that is not the whole story.
In November 2026 Claude can draw diagrams, charts, logos and icons as SVG, build interactive visuals right in the chat, design one-pagers and mockups with Claude Design, export PNG charts with code, and look at any image you upload.
This guide shows exactly what Claude can and cannot make, how to get each type of visual, and which tool to use when you need a real picture.
- No photos or AI art: Claude does not have a built-in image generation model.
- Yes to visuals built with code: diagrams, flowcharts, charts, SVG icons and logos, infographics, slides and UI mockups.
- Custom visuals (beta): interactive charts and diagrams inline in chat, on web and desktop, for all users. Download as SVG or HTML.
- Claude Design: mockups, prototypes and one-pagers on every plan, including Free.
- Claude can see images: upload photos, screenshots or charts and it will describe, read and analyse them.
- Need a real picture? Use ChatGPT Images, Gemini, Grok Imagine or a free local model in ComfyUI, and let Claude write the prompt.
Can Claude Generate Images? The Official Answer
Anthropic’s help centre article “Can Claude produce images?” says Claude does not generate photos or illustrations the way image-generation tools do. Instead, it builds diagrams, charts and interactive visuals using HTML and SVG, and it can view and analyse images you upload.
So if you type “make a photo of a cat on a beach”, you will not get a photo. You might get a simple SVG drawing of a cat, or Claude will explain that it cannot make photographs and offer a prompt you can paste into an image generator.
What Claude Can and Cannot Make
| Request | Can Claude do it? | How |
|---|---|---|
| Realistic photo | No | Use an image generator |
| Digital painting or anime art | No | Use an image generator |
| Flowchart or process diagram | Yes | Custom visuals, Mermaid or SVG |
| Bar, line or pie chart from data | Yes | Custom visuals or Python code |
| Simple logo, icon or badge | Yes | SVG code |
| Flat illustration or cartoon shapes | Basic | SVG code, quality varies |
| Infographic or one-pager | Yes | Claude Design or HTML artifact |
| Website or app mockup | Yes | Claude Design or HTML artifact |
| Slide deck | Yes | Claude Slides, export to PowerPoint or PDF |
| Edit a photo you upload (remove background, change face) | No | Use a photo editor or image model |
| Describe or read text in a photo | Yes | Upload the image |
Custom Visuals: Charts and Diagrams in Chat
Custom visuals are the newest way Claude “draws”. They are in beta and available to all Claude users, including Free, on web and desktop, in both chat and Cowork.
You do not need to turn anything on. Claude decides when a picture would explain something better, or you can ask directly.
- Open Claude on claude.ai or Claude Desktop (not the mobile app).
- Ask for a visual, for example “draw this as a diagram”, “chart this data” or “show me how this changes over time”.
- Or upload a CSV and ask “What does the data show?” to get an interactive chart.
- Interact with it: click buttons, move sliders or open it full screen.
- Keep it with “Copy as image”, “Download” (SVG or HTML) or “Save as artifact”. Visuals are not saved by default.
Anthropic’s tips: Opus does best on complex visuals, and you can set a style once (“make all my charts green”) and Claude remembers it. Custom visuals do not render in the iOS or Android apps.
How to Make Logos, Icons and Illustrations as SVG
SVG is a text-based image format. Because Claude writes code very well, it can write SVG that turns into a crisp vector graphic.
- Be specific about shape and style. Example: “Create an SVG logo for a coffee shop called Bean There: a simple coffee cup with steam, flat style, two colours (dark brown #4B2E2B and cream #F5E6CC), 512×512, no text.”
- Ask for an artifact so you see the preview next to the chat.
- Refine in plain words: “make the steam curlier”, “thicker outline”, “add the name below in a rounded font”.
- Download the SVG and open it in Figma, Inkscape or Illustrator for final tweaks.
- Convert to PNG if you need one (see the next section).
Good results: geometric logos, app icons, badges, simple mascots, patterns, map-like diagrams. Weak results: faces, hands, animals in detail, anything photo-real. If the drawing looks childish, that is the limit of hand-coded shapes, not a bug.
Claude Design: Mockups, One-Pagers and Slides
Claude Design launched on April 17, 2026 and now lives inside Claude as an artifact template. It is available on Free, Pro, Max, Ultra, Team and Enterprise plans. The old standalone site at claude.ai/design closes on December 14, 2026.
Use it for designs, interactive prototypes, one-pagers and marketing visuals.
- In any chat, ask “make a one-pager from this proposal”, or choose Output > Design in the message box.
- Or open the Artifacts tab and pick a Design template.
- Upload screenshots or a brand guide so Claude matches the look.
- Refine through chat, inline comments or direct edits on the canvas.
- Export or share by link.
Claude Slides works the same way for presentations, with export to PowerPoint or PDF. Artifacts need “Create and edit files” (code execution) turned on in Settings > Capabilities.
How to Get a PNG or JPG From Claude
Claude’s visuals start as SVG or HTML. Three easy ways to get a normal image file:
- Copy as image: on custom visuals, use “Copy as image” and paste into any app.
- Ask for Python: with file creation turned on, ask “plot this data with matplotlib and give me a PNG”. Claude runs the code in its sandbox and returns a downloadable file.
- Convert the SVG: open it in a browser and screenshot, or use Inkscape (File > Export) or any free SVG to PNG converter.
Claude Can Read Images You Upload
Claude is very good at understanding images. Upload a file with the paperclip or drag it into the chat and you can:
- Get a description or alt text for a photo.
- Read text from screenshots, receipts or handwritten notes.
- Turn a chart image back into a data table.
- Turn a hand-drawn wireframe into working HTML.
- Get feedback on a design or thumbnail.
- Check whether a picture looks AI-made (see our guide on how to tell if an image is AI).
Why Claude Does Not Generate Photos
Anthropic has never shipped an image generation model. Its focus is text, code, reasoning and agents. Image models are a separate kind of system (diffusion or autoregressive image models), and they bring extra safety work around deepfakes, copyright and non-consensual fake images.
Instead, Anthropic has leaned into visuals that are structured and editable: charts, diagrams, designs and documents. For most business use, that is often more useful than a pretty picture, because you can change every label and number.
Best Tools When You Need a Real Image
| Tool | Best for | Free option |
|---|---|---|
| ChatGPT Images | Photo-real images, text in images, edits | Limited on Free plan |
| Google Gemini | Fast image generation and edits in the Gemini app | Yes, with daily limits |
| Grok Imagine | Images and short videos | Limited |
| Midjourney | Artistic, stylised images | No |
| Local models in ComfyUI | Unlimited free generation on your own GPU | Yes, free and private |
Read more: ChatGPT Images 2.6 guide, Grok Imagine guide, is Midjourney free?, our best free AI image generators list and the best open source image models you can run locally with our ComfyUI beginner guide.
Use Claude to Write Better Image Prompts
The best workflow is to let Claude plan and another tool render.
- Tell Claude what you need: “A blog header for an article about saving money on groceries. Friendly, bright, no text.”
- Ask: “Write three detailed prompts for an image generator, with subject, setting, lighting, camera angle, style and aspect ratio 16:9.”
- Paste the best prompt into ChatGPT, Gemini or ComfyUI.
- Upload the result back to Claude and ask: “What would make this image better for a blog header?”
- Ask Claude to rewrite the prompt with those fixes.
This loop usually beats prompting the image tool alone, because Claude is good at spotting what is missing.
Troubleshooting
| Problem | Fix |
|---|---|
| Claude says it cannot create images | Ask for an SVG, diagram or chart instead, or ask for an image prompt. |
| No visual appears, only text | Rephrase: “show me a diagram” or “chart this for me”. Use web or desktop, not mobile. |
| Artifacts or downloads missing | Turn on “Create and edit files” in Settings > Capabilities. |
| SVG looks rough | Ask for simpler, flat geometric style and fewer details. Switch to Opus for complex drawings. |
| Visual gone after scrolling | Custom visuals are temporary. Use Download or Save as artifact. |
| Chart has wrong numbers | Upload the data file rather than pasting, and ask Claude to show the table it used. |
My Take
Do not pick Claude if your main goal is AI art. Pick it if you need visuals that explain something: charts, diagrams, slides, mockups and clean vector icons.
For those, it is often better than image generators, because the output is editable and the numbers are real. For pictures, pair it with a proper image model and use Claude as your prompt writer and art director.
New to Claude? Start with how to use Claude AI.
FAQ
Can Claude generate images for free?
Claude cannot generate photos on any plan. Free users can get custom visuals (charts and diagrams), SVG drawings and Claude Design mockups.
Will Claude add image generation in the future?
Anthropic has not announced an image generation model. Its visual features so far are code-based: custom visuals, artifacts and Claude Design.
Can Claude edit my photos?
No. It can analyse photos and describe changes, but it cannot change the pixels of a photo. Use a photo editor or an image model for that.
Can Claude make a logo?
Yes, a simple vector logo as SVG. Describe the shape, colours and style. For detailed or illustrated logos, use an image generator and then ask Claude to tidy the concept.
Can Claude make charts from Excel files?
Yes. Upload the file and ask for a chart. You get an interactive custom visual, or a PNG if you ask Claude to use Python.
Does Claude Code generate images?
Not by itself, but it can write code that calls an image API or a local ComfyUI server, and it can make SVGs and designs with the /design command.
Can Claude see images?
Yes. Claude can read and analyse uploaded photos, screenshots, charts and handwriting.




