Indeed. More than 300 models sounds impressive first but after stripping beta/experimental/preview & outdated/limited/redundant or painfully slow free models about 20% are left. And not all of them work immediately without customization.
It’s the lowest common denominator and adds additional costs, delays, issues and privacy concerns.
Can anybody explain the icons used in this DEVONthink 4.1 image for the OpenRouter.ai models? I want to understand if they are generated by DT4 or OR please and I note that when I see properties like this I instantly think of having a way to filter them to include/exclude just the ones I want to use. I noticed that not all properties are available on each service and that there seems to be a black or grey option for some properties. In the meantime, I’ll check the OR API and see if they have something describing all of this. Not sure if their API will let us turn on just the models that are licensed to not train on the data we send to them but it would be nice to have a filter to deactivate any LLM that doesn’t offer an appropriate privacy setting Privacy, Logging, and Data Collection | Keeping your data safe | OpenRouter | Documentation
Getting Started > AI Explained mentions the property icons.
Correction:
I found it under Settings>AI rather than Getting Started>AI
So thank you. I must have missed it before… I know I used the ? at the bottom left of the AI settings dialog but I must have been blind because it is right there now. After searching some more to find the final icon, I would probably say that if it were up to me to design the dialog for these providers, I would make the model a bit more user friendly by turning it into a table that allows filtering and priority selection. That way the user can filter out what they don’t want and select their priority for the models they wish to use if the feature isn’t available or changes and rather than icons, the titles of the table would describe the information with text, which is a lot easier to search for than images if somebody does, for example, want to understand how the $ black and $ grey are calculated by the API or the DT4 code and whether they are provided by DT4 or the provider of the LLM.
Details:
I have also found it here and it explains the magnifying glass there on page 212, but not in the DT4 docs.
DT4 Docs under Settings > AI
Using the navigation:
Using search for icon:
