🚨 Urgent: Zeabur AI Hub Image Generation API is completely down - All models and endpoints failing

Resolved

NNelson-on-codeDEVELOPER4/5/2026

Problem Description

The image generation feature of Zeabur AI Hub is not working. I have tried all models and endpoints listed in the official documentation, but all of them are failing.

Environment Information

Project Type: Node.js + TypeScript full-stack application
Deployment: Zeabur
API Endpoints:
Primary: https://hnd1.aihub.zeabur.ai/v1/images/generations (Japan)
Fallback: https://sfo1.aihub.zeabur.ai/v1/images/generations (USA)
API Key: Correctly configured in the AIHUB_API_KEY environment variable.

Models Tested

According to the official documentation, I have tried all supported image generation models, and all failed:

❌ gemini-2.5-flash-image (Marked as default/faster in docs) ❌ gemini-3-pro-image-preview (Marked as higher quality in docs) ❌ gemini-3.1-flash-image-preview (Listed on the official site)

Testing Method

My application calls the image generation API as follows:

Request Format:

const body = JSON.({
 : , 
 : ,
 : ,
 : ,
 : ,
});
 response =  (
 ,
 {
 : ,
 : {
 : ,
 : ,
 },
 body,
 }
);

Error Logs:

[ImageGen] gemini-3-pro-image-preview failed on hnd1.aihub.zeabur.ai: Model gemini-3-pro-image-preview is currently unavailable on this node (404 Not Found).

[ImageGen] gemini-3-pro-image-preview failed on sfo1.aihub.zeabur.ai: Model gemini-3-pro-image-preview is currently unavailable on this node (404 Not Found).

[ImageGen] gemini-2.5-flash-image failed on hnd1.aihub.zeabur.ai: Model gemini-2.5-flash-image is currently unavailable on this node (404 Not Found).

[ImageGen] gemini-2.5-flash-image failed on sfo1.aihub.zeabur.ai: Model gemini-2.5-flash-image is currently unavailable on this node (404 Not Found).

[ImageGen] gemini-3.1-flash-image-preview failed on hnd1.aihub.zeabur.ai: Model gemini-3.1-flash-image-preview is currently unavailable on this node (404 Not Found).

[ImageGen] gemini-3.1-flash-image-preview failed on sfo1.aihub.zeabur.ai: Model gemini-3.1-flash-image-preview is currently unavailable on this node (404 Not Found).

Code Location: https://github.com/your-username/repo-name/blob/main/server/imageGeneration.ts

Symptoms

All models return errors on both endpoints (hnd1 and sfo1).
Unable to generate any images.
Spent hours trying to troubleshoot.

Expected Behavior

According to the official documentation, at least gemini-2.5-flash-image should work and return base64 encoded image data.

Request

Confirm the current status of the Zeabur AI Hub image generation service.
Provide a list of available image generation models.
If there is a service outage, please provide an estimated time for a fix.
If this is a configuration issue, please provide the correct configuration method.

Additional Information

API Key verification is working (other AI Hub features work fine).
Confirmed model names and endpoints in the official documentation.
Tried both geographic endpoints.

2 Replies

CCanEMPLOYEE4/7/2026

Hello,

I have tested this using the AI Hub API key. The issue and solution are summarized below:

Root Cause: The Gemini image generation model on Zeabur AI Hub does not use the standard OpenAI /v1/images/generations endpoint; it must be called via /v1/chat/completions. Therefore, your original code would receive a 404 error regardless of which model you switched to.

Correct Calling Method:

const response = await fetch(
 "https://hnd1.aihub.zeabur.ai/v1/chat/completions",
 {
 method: "POST",
 headers: {
 "content-type": "application/json",
 authorization: `Bearer ${AIHUB_API_KEY}`,
 },
 body: JSON.stringify({
 model: "gemini-2.5-flash-image", // or gemini-3-pro-image-preview
 messages: [
 {
 role: "user",
 content: "Social media content frame, vertical 9:16 format..."
 }
 ],
 }),
 }
);

const data = await response.json();
// The image is in message.images (Note: non-standard OpenAI field)
const imageDataUrl = data.choices[0].message.images[0].image_url.url;
// imageDataUrl is "data:image/png;base64,....", which can be written to a file or returned to the frontend
const base64 = imageDataUrl.replace(/^data:image\/\w+;base64,/, "");

Currently supported image generation models on AI Hub (all tested and working):

gemini-2.5-flash-image (default, faster)
gemini-3-pro-image-preview (higher quality)

You do not need to change your API Key; just update the endpoint and the request/response handling as shown above. Please give it a try, and if you have any further questions, feel free to reply in this thread.

AAnonymous4/11/2026

This post has been inactive for a while. We will be closing it in 2 days if there is no new activity.

This issue has been marked as resolved

New replies are disabled for resolved issues.

Forum