How to change OpenAI Image Tools parameters?

[yevhen309]

I am experimenting with the GPT-Image-1 model.

First I tried to use it directly in chat - it doesn't seem to work that way. After reading the documentation, I set up the agent with OpenAI Image Tools and everything works fine now.
One problem was an error asking to refresh the page after image generation, but it was fixed by increasing the nginx proxy timeouts. Another problem was found - it seems (only guess) it doesn't work very well when reasoning model is used in the agent (Sonnet 4) - image is not rendered in this case. As workaround, a chain of agents can be used: one to generate a proper image description and the second to generate image itself with GPT 4.1 as a base model.

What I still can't figure out is how to change the parameters: size and quality. Is it possible to do?

Thanks in advance, thank you for this awesome project.

[jsiegmund]

Don't want to hijack your discussion but you've seemed to manage something I didn't. The instructions say you need to add the "Open AI Tools" to the tools list of the agent. Trying to add a new tool, the pop-up shown doesn't seem to contain that tool. Did that work for you? And if so may I ask which version you're running on?

[yevhen309]

Unfortunately there is nothing to hijack. For me popup shows it. I'm using the latest dev version from the main branch.

[jsiegmund]

Yeah this was my bad. I followed the instructions to update to the latest version and then the tools did pop up. Haven't gotten gpt-image-1 to work up to now so still following this one.