Inference

• All 70b parameter models have had their context lengths lengthened on both the Venice.ai app and API.

Inpainting

• Added additional inpainting blend steps per Featurebase request.

• Added graceful error handling if the inpainting pipeline was not able to determine a mask to apply from the user's input prompt.

• You can now paste an image into the chat prompt box for inpainting. Requested on Featurebase.

Token Dashboard

• Added 2 digits of precision to the Network Capacity staking numbers to avoid rounding to whole integers.

API

• Added support for an expiresAt parameter on API keys. Keys can now be set to have a future expiration date, allowing users to hand out temporary keys to others without having to remember to revoke them. These dates can be configured via the UI and via the API.

• The API docs repository is now public. For API documentation improvement, our community can now submit pull requests for review.

Characters

• Public character pages should now load faster from optomizations made to the APIs.

• If the model a public character was published with is no longer active on Venice, it will revert to the default model when editing.