Subject: Bot Acknowledges Stalling but Repeats the Same Behavior Hello Support Team, I’m experiencing a recurring issue in Chat mode. When the bot takes too long to answer, I tell it that it appears to be stalling. The bot acknowledges the problem, apologizes, and often says it w
Support Forum
@support
This group is for GizAI community support, where users can seek advice, share experiences, and solve problems together. It fosters collaboration and helps everyone stay updated on new features and best practices. If you need private support, please email support@giz.ai.
Subject: Problems With Story Control, Chat Response Times, and Image Consistency Hello Support Team, I’m experiencing several ongoing issues: Story mode: Bots frequently hijack my stories by taking control of characters, changing the plot, overriding my instructions, or disregard
Thank you for reporting these issues. We believe that the main factors behind story and consistency problems can include the difficulty of preserving long conversation context and performance limitations in models' safety and instruction-following behavior. We are not treating this as a single-cause explanation; we wil
Hi, Sorry to bother you again with chatgpt image 2. But since the issue i reported is not totally fixed, it clashs with your new politic about image generation. Indeed, i still have random number of images when i prompt. But the app considers i had it, and block the further gener
Hi A Dude — thank you for sharing this, and I’m sorry you’ve been dealing with it.
OpenAI’s current GPT Image 2 documentation confirms that prompts, input images, and generated images are subject to safety filtering. It distinguishes moderation: low (less restrictive filtering, not a bypass) from quality: low (fas
Hi fanfan, thank you for reporting this, and we’re sorry.
You’re right: OpenAI’s response can sometimes contain fewer images than requested. Our app could still charge your quota for the full request, which then blocked further generations. We’re sorry—we’ve confirmed the issue and are preparing a fix. We plan to depl

What's this? Am I going to be charged for using GPT Image 2 now? Not that it's actually that useful anymore anyway.
Thanks for flagging this, and sorry for the confusing display. You were not charged for the GPT Image 2 generations shown here. GPT Image 2 is included with your Plus plan for up to 10 uses per hour; credits are used only if you choose to continue after that allowance. We’ve updated the settings to show the included al

I'm getting an error that says "All provider routes are temporarily unavailable. Please try again shortly" regardless of the model.
Thanks for reporting this, and sorry for the interruption. This was a brief upstream provider outage and was not related to your account. Service is back to normal, and we’ve added additional backup provider routes to reduce the chance of the same interruption affecting multiple models again. Please try your request ag
400 status when sending prompts to AI Chat :( (all the models I've tried)
Hi MrsQueenSize, sorry about this, and thank you for reporting it. We found an issue in a shared AI Chat request step, which is why it could affect every model you tried. The fix is now live, and we have verified successful AI Chat responses in production. Please refresh the page and try your prompt again. If you still

I see no "Put the characters in background" mode in ltx distilled 2.5 - will it ever be added again? And if not, what's the alternative. Also the default mode is failing all the time, but it might be because of migration from 2.3 to 2.5
Thanks again for reporting this, and sorry for the trouble.
We have fixed the issue that was causing LTX-2.5 Standard generations to fail. The closest supported alternative to the former “Place Subjects in a Background” workflow is Ingredients: upload one reference sheet containing both the subjects and the background
Thanks for reporting this, and sorry for the confusion.
The “Place Subjects in a Background (MSR 2.0)” workflow is not currently available in LTX-2.5. We’ll consider reintroducing MSR once the underlying project update is available and the workflow is compatible again.
For now, the closest alternative is the Ingredie
Hi, Just to say that 90% of my generated images cross all the conversations i did just vanished Is it normal ? Thanks
Hi, this was not normal — sorry about that, and thank you for reporting it. We found and resolved an issue that temporarily prevented some images from older generations from loading in past conversations. Please refresh GizAI and reopen those conversations; the images should now be visible again. If any older images ar
error messages all day about outputformat is required. or no size chosen and I have all boxes checked.
Hi Trevor,
We’re sorry for the errors you ran into. They were caused by a regression in a deployment about an hour ago, and we’ve now fixed it.
Please refresh GizAI and try again. Thank you for reporting this and for your patience.
GizAI Support
I’m uploading JPG images from a USB thumb drive. Some JPG files upload successfully, but many others produce the error: “Trash files cannot have public media links.” The affected files are stored normally on the drive and are not in a Trash or Recently Deleted folder. This appear
Thanks for the detailed report, and sorry for the trouble. We found the cause: if a file is uploaded with the same name as a file previously moved to Trash, its storage record can remain marked as Trash and block the public media link needed for generation. We have prepared a fix, but it is not live in production yet.

I keep running into issues where GPT Image 2 will generate an image. I try it again, and, still using the same prompt that just worked a second ago, all of a sudden it trips the safety censors for some reason. Or it stalls at a time limit and sits there.

One day, the developers behind AI are going to realize that they're no more responsible for what people do with the tools they created than Bic is if someone uses one of their pens to write a naughty word on a wall someplace.
But it is not this day, apparently.

I think the problem here is that OpenAI recently changed their classifiers. I ran some tests via the API and can confirm that 9 out of 10 prompts that worked fine before now trigger sexual moderation, even if the prompt specifies that the body is hidden or covered. Doing mostly pinup/pulp style, this is a big bummer fo
Thank you for the update. I appreciate the model change and the clarification that Story continuity and video generation are being investigated separately. I’ll test the updated Assistant model and monitor whether the instruction-following and back-and-forth issues improve. Howev
We don’t have a confirmed timeline yet for Story continuity or video-generation consistency improvements. These remain separate issues under investigation. If either problem recurs, please share the conversation link or approximate time and, for video issues, the selected video model. Screenshots are not required; this