Anthropic has rolled out a new beta feature for its AI assistant Claude that lets it generate interactive charts, diagrams, and data visualizations right inside chat conversations. Instead of answering every question with plain text, Claude can now build visual explanations on the fly — and update them as the conversation continues.
The feature is available to all Claude users, including those on the free plan, and is enabled by default on web and desktop apps.
How Claude’s new visualization feature works
Anthropic describes this capability as an extension of “Imagine with Claude,” an experimental project the company previewed in fall 2025. That prototype let Claude subscribers create custom user interfaces on a virtual desktop in real time without writing code. The new visualization feature works on a similar principle, but it is integrated directly into the chat experience.
What is important to understand here is that Claude is not generating images in the traditional sense. According to Anthropic’s support documentation, these visuals are built using HTML code and SVG vector graphics — essentially the same building blocks used for web pages. Think of it as Claude having access to its own whiteboard that it draws on while explaining something to you.
These inline visuals are different from Claude’s existing Artifacts feature. Artifacts are persistent, shareable documents and tools that appear in a side panel and can be downloaded or published. The new visualizations, on the other hand, are temporary. They appear directly within the conversation flow, change as you ask follow-up questions, and may disappear as the chat evolves. However, you do have the option to save a visual as an SVG or HTML file, or convert it into a full artifact if you want to keep it.
What can Claude visualize?
Anthropic has shared a few examples that give a good idea of what this feature can do. Ask Claude how compound interest works, and it will generate an interactive growth curve where you can adjust parameters like principal, rate, and time to see how the numbers change. Ask about the periodic table, and it will create a clickable, interactive table where you can tap on any element for more details.
Beyond educational content, Claude can also generate flowcharts to explain how a process works, structural diagrams showing weight distribution in a building, step-by-step visual guides like paper airplane folding instructions, and data charts for business use. For example, if you are an entrepreneur, you could feed Claude your revenue data and ask it to chart your monthly or quarterly trends — a useful way to visualize business performance without firing up a spreadsheet tool.
You can ask for visuals or let Claude decide
Claude will automatically determine when a visual response would be more helpful than plain text. But you can also request one directly using prompts like “draw this as a diagram,” “chart this data,” or “visualize how this might change over time.” Once a visual appears in the chat, you can interact with it — click buttons, adjust sliders, or expand it to full screen — and then ask Claude to refine it further or add more detail.
Availability and current limitations
The feature is live now in beta across all Claude plans, including the free tier. However, there are a few things to keep in mind. Custom visuals currently only render on Claude’s web and desktop apps. If you are using Claude on iOS or Android, you will not see the visual elements — Claude will still answer your question, but in text form only.
Anthropic also notes that visual quality and complexity can vary since this is still a beta release. For more complex visualizations, the company recommends using the Opus model, which handles visualization tasks better than Sonnet or Haiku.
Part of a broader push for richer responses
This launch does not exist in a vacuum. Earlier this year, Claude started using purpose-designed formats for specific types of content — recipes now appear as formatted cards with adjustable servings, weather queries show visual forecasts, and Claude can present structured multiple-choice questions instead of requiring typed responses. These visuals can also interact with third-party apps like Figma, Canva, and Slack through Claude’s connectors.
The timing is also notable. OpenAI introduced a similar feature for ChatGPT just days earlier, adding interactive visualizations for math and science concepts. The parallel moves suggest that inline visual content is becoming a key battleground in the AI assistant space, as companies push beyond text-only interfaces to make AI responses more intuitive and useful.