Image gen not supported/enabled for gemini-3-pro-image-preview?

[JoeGrimes123]

Describe the bug
Can't use gemini-3-pro-image-preview or gemini-2.5-flash-image from Antigravity to provide any image.

Provider
Antigravity

Model Name
gemini-3-pro-image-preview , gemini-2.5-flash-image

Request Information
Cherry Studio chat

Expected behavior
Cherry Studio should be able to detect that the model supports vision/imagen

Screenshots

OS Type
Windows

Additional context
Also idk how to use curl or create script to make it use imagen tool. Tried both code sample from gemini docs and openai docs but didn't work

[luispater]

Please don't use cherry.

It has a lot of bugs.

You can try gemini-*-image on NB

[JoeGrimes123]

Cherry, LM studios Msty studio, other LLM chat testers are all not able to generate any image using the openai compatible schema so I don't think its a bug but I know Antigravity tool/function calling is troublesome
The nano-banana image generator link you gave keeps giving API error 404: 404 page not found. even tried asking AI to recreate code to a simple js from the source but still wont work. Could you please share the basic code for image gen? I'm too dumb to understand anything

[luispater]

You should use an endpoint like http://127.0.0.1:8317/v1/chat/completions

Almost all LLM chat clients do not support image generation.

If you want to use OpenAI Chat Completions protocol, you should follow this document image-generation, because OpenAI doesn't define the image generation format in the Chat Completions protocol.

[JoeGrimes123]

tqsm luis, I got it working. Openrouter docs rocks!
here's the sample code I used:

TS code

// @ts-ignore
import * as fs from 'fs';
import { Buffer } from 'buffer';

const response = await fetch('http://127.0.0.1:8317/v1/chat/completions', {
  method: 'POST',
  headers: {
    Authorization: `Bearer api-key`,
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({
    model: 'gemini-3-pro-image-preview', // gemini-2.5-flash-image, gemini-3-pro-image-preview
    messages: [
      {
        role: 'user',
        content: 'Create a high quality realistic picture of hybrid animals between a monkey and a panda.', // prompt
      },
    ],
    modalities: ['image', 'text'],
    image_config: {
      aspect_ratio: '16:9',
      image_size: '2K', // 1K, 2K or 4K
    },
  }),
});

const result = await response.json();

if (result.choices) {
  const message = result.choices[0].message;
  if (message.images) {
    message.images.forEach((image: any, index: number) => {
      const imageUrl = image.image_url.url;
      console.log(`Generated image ${index + 1}: ${imageUrl.substring(0, 50)}...`);

      const base64Data = imageUrl.replace(/^data:image\/\w+;base64,/, "");
      const buffer = Buffer.from(base64Data, 'base64');
      fs.writeFileSync(`./image.png`, buffer);
      console.log(`Saved image to ./image1.png`);
    });
  }
}

export {};

Py code

import requests
import json
import base64

url = "http://127.0.0.1:8317/v1/chat/completions"
headers = {
    "Authorization": f"Bearer api-key",
    "Content-Type": "application/json"
}

payload = {
    "model": "gemini-2.5-flash-image",
    "messages": [
        {
            "role": "user",
            "content": "Create a high quality realistic picture of hybrid animals between monkey and a panda."
        }
    ],
    "modalities": ["image", "text"],
    "image_config": {
        "aspect_ratio": "16:9",
    }
}

response = requests.post(url, headers=headers, json=payload)
result = response.json()

if result.get("choices"):
    message = result["choices"][0]["message"]
    if message.get("images"):
        for image in message["images"]:
            image_url = image["image_url"]["url"]
            print(f"Generated image: {image_url[:50]}...")

            with open("image1.png", "wb") as f:
                if image_url.startswith("data:"):
                    # Handle base64 encoded image
                    header, encoded = image_url.split(",", 1)
                    data = base64.b64decode(encoded)
                    f.write(data)
                else:
                    # Handle regular URL
                    f.write(requests.get(image_url).content)
            print("Saved image to image1.png")

[luispater]

That's great. Image generation is not defined in the OpenAI Chat Completions protocol, and OpenAI no longer maintains Chat Completions.

So, the OpenRouter image generation documentation has essentially become the standard, and we have specifically implemented compatibility for it.

[LuckyGJX]

Why did I get this error，use python

'error': {'code': 400,
'message': 'Multi-modal output is not supported.',
'status': 'INVALID_ARGUMENT'}}