I wrote an email to TheBrain's support team yesterday inquiring about how AI API works in this app. Essentially, I praised them for the app itself (which is fantastic), but I expressed how the only thing that keeps me from using its AI chat (Cerebros) is how it seemingly only supports OpenAI's models. I genuinely don't want OpenAI and its models to meddle with my projects, so I inquired about the possibility of supporting local models or API keys from other services, companies.. (with OpenRouter being my biggest recommendation).

Steven Rico (Part of TheBrain Support, God bless his soul) replied to me essentially saying that local model support ALREADY exists (which is fantastic, I didn't know that) and gave me the instructions on how to use them, but TheBrain indeed only accepts an endpoint URL with a model's name.. so no custom API key (and therefore OpenRouter and services akin to it won't work directly with the app), and the only way this can be used is through a custom setup to bridge the gap (aka an experimental feature that may or may not work, and TheBrain won't provide setup support for (which is completely reasonable)).

He did however point out that I could file a request about it on UserVoice.. and so here I am. If you guys could create some kind of "Global API Compatibility" layer by default (rather than OpenAI), that would be INSANE. People would be able to use all kinds of models and services of all varieties (free or not when it comes to tokens), and that would turn TheBrain into a personalized agentic harness pretty much (and a powerful one at that).. ON TOP of everything it already does (so imagine the potential). This is an extremely useful and important update for people like me who're not just cautious with their data and privacy, but also conscious with their money (as there's countless alternatives to GPT models that're MUCH cheaper, on top of being as good (some even better) and don't mess with your privacy).. on top of the fact that this is also an important global compatibility thing (OpenRouter gives access to over 500+ models at absurdly cheap prices, so this is also about letting people use whatever models they want/prefer), and this gives people an infinite amount of alternative choices rather than just having to run models locally when trying to run away from Sam Altman and his shady license terms and pricing (which is not ideal since the large majority of people don't have enough hardware to handle big models).