More choice over our models
We operate models on infrastructure we control instead of making the experience dependent on a single external model API. That approach gives us choices about deploying, adapting and replacing models across conversations and creative media. It concerns how we run the service, not a promise that a particular model will remain unchanged for the lifetime of every Kyn.
Independence has limits
Control over deployment can reduce reliance on another provider's model availability or content policy. It cannot guarantee that every response will be right, every generation will be immediate or access will never be interrupted. Privacy also depends on the way information is handled and the preferences available in your account. Infrastructure is one part of that responsibility, not a replacement for it.
Keep character and model changes distinct
Operating our models requires maintenance, capacity planning and incident response. Interruptions remain possible, and the models themselves can change. A continuing character identity does not freeze every underlying feature forever. Our account terms, privacy choices, feature guidance and release notes explain different parts of that relationship as the service develops.