I am currently running an API server. However, it does not support automagical AI model detection. Can DT add the support of the entry of model name manually? Thanks!
The endpoint /v1/models isn’t supported or does it have a different URL? The model list is also required to figure out the supported capabilities of each model.
I use an API proxy called Close-AI. They provide a variety of models for users, and users need to enter the model ID themselves in order to use a model.
Here is their documentation:
I suddenly ran into the same problem as well. After switching to another third-party API provider, I could no longer see the model list. Even after switching back to the original provider, the list still wouldn’t load. I also tried other AI apps that need to load models, such as Cherry Studio, and they worked fine.
Which provider and how many models are provided? DEVONthink probes each model first, this might require some time depending on the number of models.
The new one is Poixe AI and old one is https://openkey.cloud. There’re more than 100 models for each provider.
Try restarting DEVONthink and then waiting for some time. In addition, is the entered URL the right one (the chat completions endpoint)?
Currently, my AI-powered Smart Actions can still run. I assume that’s because the required models had already been configured, and I haven’t changed those settings. However, the model list is still not visible in the Settings panel.
Please choose Help > Report Bug while pressing the Option modifier key to start a support ticket so that we can have a look at the logs.
Emphasis mine
Thanks!