I’m not sure what’s happening, but it looks like it’s defaulting to the wrong LLM. How can I tell what’s really going on here and why this is happening? It’s a significant performance decrease when it’s using the wrong LLM.
Hi @chrisbehnke ,
I investigated this and found the affected agent. Luna is failing because the connected OpenAI API key has no credits remaining.
Because the workspace’s fallback model setting was enabled, Pickaxe automatically retried those requests using GPT-5.1 and Pickaxe credits. After I disabled fallback temporarily, OpenAI surfaced the underlying billing error shown in the screenshot.
Please add credits to the OpenAI organization associated with that API key. Luna should then work normally again.
For future reports, please include the link to the affected agent in your post or email it to us. That helps us investigate much faster.