AI Hub is becoming the core layer for model access across vendors. That means two things:
- we keep the catalog current as vendors deprecate models, and
- we keep expanding what you can build through a single interface.
Alongside that, we’re shipping meaningful improvements to our flagship editing services, especially for high-volume, real-world workflows like automotive marketplaces.
AI Hub Updates
Catalog maintenance: deprecated models removed
Some image-generation models were decommissioned by Google and have been removed from Picsart AI Hub.
If you pinned a decommissioned model, please update your integration to a supported option. If you don’t pin a model, requests may still be handled gracefully by routing to an available default—but for predictable results, we strongly recommend explicitly selecting the model you want to run.
👉 AI Hub overview
Text2Text added: more models, more vendors
AI Hub now supports Text2Text with 21 models across 3 providers:
This makes it easier to standardize text workflows (prompting, rewriting, classification, planning) without stitching together separate vendor integrations.
👉 Learn more
Video generation added to AI Hub
We launched Text-to-Video and Image-to-Video this month, and AI Hub now includes 51 video generation models across 12 providers.
Instead of building model-by-model, you can explore options through one consistent surface area, and keep your integration stable as the ecosystem evolves.
👉 Explore video generation services:
These services are designed to fit into automated workflows. They work well for creative iteration, marketing variations, and rapid experimentation, especially when combined with AI Hub for model flexibility.
Smart Crop: more control with sizing parameters
Smart Crop now supports new parameters for tighter control over output:
widthheightratio
This makes Smart Crop easier to use in production pipelines where output sizing must match specific placements (product cards, ads, templates, marketplaces).
Remove Background: upgraded for car photos
Remove Background is a flagship service for many customers. This month we shipped quality improvements specifically tuned for vehicle photography.
If you manage automotive inventory, marketplaces, or dealership feeds, you should see more reliable cutouts on challenging frames and complex car shapes.
Video APIs: better compositing and better audio workflows
Video Remove Background: transparent and semi-transparent backgrounds
Video Remove Background now supports transparent and semi-transparent backgrounds, making compositing workflows smoother and more natural.
Transcription outputs expanded: SRT, VTT, SBV
Transcribe Audio/Video now supports:
- SRT
- VTT
- SBV
You can also choose 1-word, 2-words, or sentence granularity, depending on your subtitle and editing workflow.
Extract Audio from Video
A new service to extract the audio track from a video file.
👉 Extract Audio from Video API
Looking for the full change log?
👉 Browse the release notes
We’re just getting started.
Stay tuned for more updates and see you in February with the next wave of improvements.