Frequently Asked Questions

What is GLMImage?

GLMImage is an independently operated AI image creation product. It supports text-to-image and image-to-image workflows. It is not an official Zhipu AI product and is not endorsed by Zhipu AI.

Does every request use the GLM-Image model?

No. GLMImage is the product name. Production requests may be routed to supported models according to the workflow and current routing configuration. The provider and model stored on each task are the source of truth for that generation.

Do I need to sign in?

Yes. Sign-in is required before a generation task can be created. If you start from the generator as a guest, GLMImage preserves the current draft through the authentication flow.

Is there a free trial?

Yes. Every new account receives 20 credits, enough for 2 trial images (each generation costs 10 credits). After the trial credits are spent, additional generation requires credits from a subscription or credit pack.

How many credits does an image use?

The current image generation cost is 10 credits per image. The interface shows the total credit cost before a task is submitted.

How long does generation take?

Timing varies by model, provider capacity, image settings, and current service load. The task status is shown while processing; GLMImage does not promise a fixed completion time.

Which output sizes are accepted?

Width and height must each be between 256 and 1536 pixels and divisible by 32. The interface provides validated presets and checks custom dimensions before submission.

How do reference images work?

Image-to-image workflows require at least one uploaded reference. A workflow may allow up to four references. The interface displays the applicable minimum and maximum before generation.

Are my reference images published?

No. Reference images are not included when a result is submitted to the public Gallery. A generated result is only submitted when you explicitly choose to publish it.

Can generated text contain mistakes?

Yes. In-image text quality varies by the routed model. Review every word, especially small text, before publishing. Use a layout tool for long copy, legal text, or QR codes.

What should I do if task creation fails?

Check the prompt, output size, reference count, authentication state, and credit balance. GLMImage shows a categorical error without exposing a provider's raw error response. Retry after correcting the setting, or contact support if the same category continues.

Where can I see the task model and status?

Open Activity and select the generation task. The stored task record includes its status, provider, model, scene, and credit cost.