Hey, thanks! I appreciate that!
Right now we're using a mixture of third party models (mostly OpenAI and Mistral models) depending on the task with some extra work on our end to augment prompts, do failover when someone's API goes down, etc, but we're also planning on training our own models for more specific tasks that also include game state (dialogue, decisions, etc) with some options to pick and choose which models you want for specific API calls or tasks. We're building it for our own games as we go so the plan a little hand-wavy right now, but I was hoping the survey would give us a better sense of the kinds of features other people might want/need as well. We're having a ton of fun with the internal API right now and we figured it's the kind of thing that other people might get a lot of use out of, too.
Thanks again for filling out the survey, we're getting a few responses slowly and that's exciting -- any responses at all are better than none! :)