Summary
- Snapchat is developing an in-house AI image generation model.
- The AI model can produce high-resolution images in under 1.5 seconds (depending on your device).
- Snap plans to implement the AI model in Snapchat features later this year.
Snapchat is getting in-house AI image generation. As reported by TechCrunch, Snap has announced a new text-to-image model that’ll be used in a variety of Snapchat features. The company says that the on-device model is capable of producing high-resolution images in under one and a half seconds, using an iPhone 16 Pro Max.
Snap says that it developed its own AI model, which is “capable of generating high-resolution images for mobile devices in just seconds.” This new on-device capability will power “AI Snaps, AI Bitmoji Backgrounds, and more,” and Snap expects to bring it online this year.
As TechCrunch points out, Snapchat has previously leveraged third-party AI to power certain features. For example, Google Gemini has powered both a Lens-style visual search feature and Snapchat’s My AI chatbot.
Coming soon to a Snapchat near you
Snap says it wants to get Snapchat’s homespun AI model up and running “in the coming months.” Given it’s an on-device process, how well it’ll work for you depends on which phone you’re packing. If you’ve got a high-end phone (like the aforementioned iPhone 16 Pro Max), you can probably expect results within a couple of seconds. More entry-level models will take longer to perform the same generative tasks, if they’re compatible at all.
