Using Gemini Nano Banana AI, I gave it a command, but it replied:
“This is a great task! To create an attractive food-festival promo DM, the key is ‘mouth-watering visuals’ and ‘clear information.’
Since I am an AI text model, I cannot directly output a final image file (like JPG or PDF). However, I can provide a full DM layout structure, copywriting content, and visual guidelines for designers (or for you when using Canva).”
But… Isn’t Gemini Nano Banana AI supposed to create images?
Such a huge bug — does the development team really not fix it?
Hi @BearBaby
Thank you for reaching out to us.
If possible, could you share the exact prompt you are trying to run? Also, could you please confirm which platform you are using?
This will help us better understand and analyze the problem so we can provide you with a more accurate and helpful response.
@**Shivam_Singh2
use chrome and firefox **
On 12/01, I activated the PRO plan.
From 12/02 to 12/03, the issue occurred multiple times: I was unable to generate images (even when selecting Gemini Nano Banana AI – Image Mode).
I was also forced back into text-only mode, and the system responded saying Gemini is only a text model and cannot generate images.
(Please refer to the image.)
Status:
Already switched to Gemini Nano Banana AI mode and Thinking Mode.
Command:
Create a promotional DM.
Event Date: 2025/12/06 (Sat)
Event Time: 14:30–16:30
Event Location:
Event Highlights:
Event Schedule:
14:00–14:30 Music Warm-up
14:30–14:40 Arts Performance
14:40–15:00 Opening Ceremony
15:00–15:50 Themed Food Tasting
15:50–16:00 Arts Performance
16:00–16:30 Singing or Dance Performance
@Shivam_Singh2
This screenshot shows the exact situation I encountered and the reply I received when I asked Gemini directly about the issue.
And here’s the part that is absolutely unacceptable:
Even Gemini itself responded in a way that clearly indicates the situation is nonsensical — essentially acknowledging that this is a serious structural failure of the system.
This is not a random glitch, and definitely not a one-time mistake.
A problem of this scale should have been caught, prevented, and resolved during the testing phase, long before being released to users.
Yet it keeps happening. Repeatedly.
There is absolutely no justification for a malfunction this fundamental to still exist.
