GPT-persona, Stability and Dall-E

AI-generated summary

Jagster is experiencing issues with their GPT-persona and image generation after upgrading. Initially, they were using the AI Stability API key, but after an upgrade, they found that their personas could no longer generate images due to unsupported languages and a lack of settings. After deleting the Stability API key, they found that their personas could only use Dall-E for image generation. Jagster also noticed that the “Tools that a persona can use” section was missing the Dall-E option, which seemed to be the cause of the issue. However, after resolving the issue, they found that they could no longer specify direction, resolution, or use non-English languages like Finnish, which was possible before with Dall-E.

I’m now quite lost.

I have/had API-key of AI Stability in use, and otherwise I’m using OpenAI.

When models started use Vision I found that I can ask images using what ever persona, even prompting in Finnish (generally using something else than English for system and image prompts is bad idea).

I upgraded yesterday and few moments ago I started play with one of my ”general” GPT-persona and images. Results were… low quality is pollite expression.

I didn’t totally understand what heck happened, but when I prompted an image in Finnish I got error. Un-supported language by Stability.

That explained a lot like why I suddenly started get four images instead one.

I deleted Stability API and tried again. Now my personas can’t create images any more, except Dall-E persona of course.

So my question is

  • is something changed
  • is it been that way all the time
  • have I missed some setting, again

This isn’t my day.

Tools that a persona can use misses one small thing: dall-e.

Still something has changed. Earlier I could tell direction, resolution and even use Finnish. Because Stability doesn’t allow (or follow) those a persona has used Dall-E — without that tool.

What ever, now it works.