Google Pics brings object-level AI image editing to Workspace
Built on Nano Banana: object edits, in-image translation, and 4K upscale from pics.new — but no API yet. Plus Perplexity's on-device privacy gate for Mac.
Copy markdown
Object edits, in-image translation, 4K upscale
Pics selects and re-edits individual objects in an image, rewrites or translates the text baked into a graphic, reformats for web/social/print, and upscales to 2K or 4K — with batch edits you can revert one by one. It runs as an overlay inside Google Docs and Slides, or standalone at pics.new. For anyone shipping marketing or social assets, the in-image translation alone turns localizing a graphic into a two-click job.
The catch: it's a UI, not an API
Pics is live now for Google AI Pro and Ultra, Workspace Business Standard/Plus, Enterprise Standard/Plus, and AI Pro for Education — but there's no public API. You can't wire Nano Banana's object-level editing into your own pipeline yet, so for now it competes with Canva, not with an image model you'd call from code.
Perplexity adds an on-device privacy gate for Mac
Perplexity's new Hybrid Compute classifies every prompt and keeps the sensitive ones on a local model running on Apple Silicon, sending everything else to cloud frontier models — with no token charges for the local runs. It's a clean, shippable pattern for keeping regulated or private context off the wire while still reaching for a big model on the rest.
Elsewhere: a coding-tuned Gemini 3.8 Flash reportedly imminent
Multiple outlets report Google is prepping Gemini 3.8 Flash — a cheaper, coding-focused Flash — to land as soon as this week, roughly two weeks after 3.7 Flash. There's no official model page, pricing, or confirmation yet, so treat it as a heads-up to watch, not something you can call today.