
Claude can analyze images but not create them. How MCP connectors fix that, and the two-minute setup that adds real image models to Claude.

What it cannot do natively: render a photo, logo, illustration or any other raster image inside the chat.
Claude supports MCP connectors: external tool servers that plug new capabilities into the chat. Krater ships one, and adding it is the shortest path from "Claude can't draw" to "Claude just made this image".
Then ask: "generate a product photo of a ceramic mug on a linen tablecloth, soft morning light" and Claude calls Krater's image tool with the right model for the job. Because the tools run on your Krater plan, one subscription covers images, video, voiceovers and documents, not just pictures. See the full Krater MCP setup.
The direct route: generate inside Krater itself. The image studio has the same models (Flux, Imagen, Ideogram, Recraft and more) behind one prompt box, and the chat assistant can route an image request to the right tool automatically. No connectors, no MCP, no config: krater.ai/image.
Claude itself has no image generator at any price. With a connected tool like Krater MCP, image generation runs on that tool's plan, not Anthropic's.
Not natively. It can draft SVG code for simple graphics, but for real logos, illustrations or photos you need an image model, which is what a connector adds.
No. Anthropic does not bundle any image model. MCP connectors are how third-party generators like Krater's plug in.
It can analyze and describe it, and suggest edits, but it cannot output an edited image file without a connected tool.