Is there any way to use image generation models available on Shuttle AI?

[kneelesh48]

The chat models work, but selecting image generation models from the drop-down and sending a chat message results in api error as shown in the screenshot below

[fuegovic]

Maybe, but not easily (this has not been tested and might not work at all)

Image generation only works with plugins currently. You might be able to send a request to the dall-e / dall-e-3 models if you configure the following .env variables to point to ShuttleAI:

You can try setting either

DALLE_REVERSE_PROXY=

or:

DALLE3_BASEURL=
DALLE2_BASEURL=

Then install the DALL-E and/or DALL-E-3 plugins and test

Another thing to consider is that the plugin endpoint is using OpenAI for the thought process. If you don't have OpenAI access you would also need to configure ShuttleAI to hijack the OpenAI endpoint with this .env variable:

OPENAI_REVERSE_PROXY=

Even after completing all these steps, success is not guaranteed and the models would be limited to dall-e and dall-e-3.

---

@danny-avila
Would it be possible to introduce endpoints dedicated solely to image generation in LibreChat? Although we currently support numerous endpoints capable of image generation, there isn’t a straightforward method to use the feature.