AI image generation inside ClawAI conversations

Several providers behind one request

Image requests can be served by Gemini image models, OpenAI’s gpt-image-1, xAI’s Grok Imagine, or local Stable Diffusion models (SDXL-Turbo and a ComfyUI workflow) running on your own hardware. If the provider that was chosen fails, the request moves to the next one — cloud first, then local — instead of returning an error.

If you pick a specific image model yourself, that model is used, even when the prompt does not contain an obvious image keyword.

Generated in the conversation, with progress

Images appear inside the chat with a progress panel while they are generated. You can cancel a generation, or retry with a different provider if you do not like the result. There is no separate image app to switch to.

Prompts, sizes and edits

Prompts can be up to 4,000 characters, and images can be requested at sizes from 256 to 4,096 pixels. Attach a reference image of up to 25 MB to edit an existing picture rather than start from scratch. Image generation is a paid-plan feature and is metered on its own surface, separate from chat tokens.

Questions people ask

Which models generate the images?

Gemini image models, OpenAI’s gpt-image-1, xAI’s Grok Imagine, and local Stable Diffusion models. Which of them are available depends on what the operator has configured.

Can I edit an image I already have?

Yes. Attach a reference image of up to 25 MB and describe the change you want, and the model edits it instead of generating from scratch.

Is image generation available on the free plan?

Image generation is a paid-plan feature and is metered separately from chat. The pricing page shows which plans include it.