April 16th, 2025
Previously, auto mode would sometimes classify character dialog or narrative actions as OOC. Accuracy has been improved, you should see less messages classified incorrectly.
Weโve made various smaller tweaks to the prompts. Please let us know if you feel anything different about the changes. I know it might feel strange we arenโt telling you specifically what is different, but prompting/quality of outputs can be very subjective. We donโt want to bias your feedback one way or another. Getting your unbiased feedback helps us determine if our changes were good or need to be reverted.