Currently summaries are generated by gpt-5-nano which creates poor, almost unusable summaries. I understand the financial risk of supporting a better model since NoteCat allows for unlimited usage. As a compromise, I suggest that users be allowed to provide their own OpenAI API keys and given the ability to select higher-tier models. This way as the user I can decide whether I want to accept the cost tradeoff for bleeding-edge models like gpt-5.6-sol without making the service more expensive for everyone.