Vlad I'm not sure that I personally would pay $10 extra just for image generation, since I don't use stand-alone image generation a ton in my workflows. But there are definitely situations where image generation in a search or search-adjacent workflow would be really helpful.
For example, if I were using Research Assistant and had already uploaded an image, pulling in an image generation model like Stable Diffusion would be a natural fit at that point and would make a lot of sense, and I would pay for the convenience of being able to do that.
Also, I find that I typically switch to Google to do image search, so having the option to load AI-generated image as part of an image search workflow would be a real differentiator if none of the Kagi search-provided results fit what I'm looking for.
So I guess that has me leaning more towards a pay-per-use model, at least at first, unless that $10/m had a variety of different models to meet different use cases (that would also help reduce adverse selection, e.g. a user upgrading just for a month to use image generation a ton and then immediately downgrading, if there is a bundle of various models/applications, that could lower churn).
Then again, I didn't think I would need Kagi Ultimate, but now that I have it, I can't imagine life without it 🤷
I definitely agree with Browsing6853 that it shouldn't be forced onto the larger Kagi user base as a price hike given the animosity some users have towards AI "bloat." FWIW I too dislike AI bloat, but my sense is that the Kagi team has a good sense for where the line is (Exhibit A), and the decisions made so far have engendered my trust.
