Missing jupyter-ai models using langchain-aws

In a JupyterLab server I’m using jupyter-ai==2.31.7 and langchain-aws==0.2.35. When looking through the Bedrock model options, the most recent ones like Sonnet 4 and Opus 4.5 are missing. How can I go about enabling them in jupyter-ai?

I completely destroyed a TLJH instance yesterday, while trying to do that via loading a beta 3.0 jupyter-ai, and using several suggestions from AI. When it arrives and is fully supported, you might be able to enter your model, rather than just use a drop down in the AI chat window. Before I destroyed TLJH, it was possible to get anthropic:sonnet-4-5-??? to work using the ai magics. By that time, the AI chat window was completely empty. :slightly_frowning_face:

I had been using gemini 2.5, but was not particularly impressed with it. I expected better results from upgrading to sonnet 4.5. It worked well where I had used it before, but the platform was unpleasant to use.

Overnight, I was wondering about finding the source with the dropdowns and changing them to match what I wanted. I know I might likely need to track down URLs and change them to match. No idea, whether the API changed, but in two years that seems to be a likely possibility.

I likely may need to dig through the extension source to understand how langchain-aws is used to populate the dropdown list.