search

enricoros / big-AGI

Generative AI suite powered by state-of-the-art models and providing advanced AI/AGI functions. It features AI personas, AGI functions, multi-model chats, text-to-image, voice, response streaming, code highlighting and execution, PDF import, presets for developers, much more. Deploy on-prem or in the cloud.
https://big-agi.com
MIT License
5.33k stars 1.2k forks source link

[BUG] Azure OpenAI DALL-E 3 not displaying #452

Open pascalvenema opened 6 months ago

pascalvenema commented 6 months ago

Description

Even though I have deployed DALL-E 3 as a model in Azure OpenAI, and can access it through the playground, it is not displaying as a model in big-AGI.

Things I have tried:

Device and browser

MacBook Air - Microsoft Edge

Screenshots and more

Screenshot 2024-03-11 at 17 37 42

Willingness to Contribute

enricoros commented 6 months ago

Hi @pascal1888 I didn't know that Dall-E models were available in Azure. Which geo/data center do you use to get DallE models available?

Cozy-T commented 6 months ago

Hi @pascal1888 I didn't know that Dall-E models were available in Azure. Which geo/data center do you use to get DallE models available?

Hi @enricoros, currently it's East US, Australia East and Sweden Central.

You can look it up here: https://learn.microsoft.com/en-us/azure/ai-services/openai/concepts/models#dall-e-models

enricoros commented 6 months ago

Thanks. The Text2Image subsystem is being improved to support LocalAI and other providers with dynamic (e.g. not known a priori) image generation models. I may have Azure as part of the dynamic discovery at the same time.

clawfire commented 3 months ago

Hey there. Azure openAI provide Dall·E to more clients on Sweden Central know. Would be awesome to be able to use it, especially for people with constraint on privacy regarding openAI classic flavor 😃

(I want to take that opportunity also to say how AMAZING BIG-AGI is 😄)

enricoros commented 3 months ago

Thanks @clawfire . In principle it's quite easy to do, most of the infrastructure is in place already. I'll mark this as an easy ticket for first configurators, as I would do it, but it's below in priority vs the larger multimodal work I'm doing at the moment.