Google's Nano Banana 2 Lite Makes AI Images in Four Seconds
Google's cheapest image model yet costs $0.034 per 1,000 pictures and, on the company's own benchmark, edges out its pricier sibling.
Four seconds, and a fraction of a cent. Those are the two numbers Google wants developers to remember about its newest image model, and they say more about where generative media is heading than any demo reel.
On Tuesday the company released Nano Banana 2 Lite, the stripped-down, high-speed sibling of its flagship image generator, formally called Gemini 3.1 Flash-Lite Image. It produces a picture in about four seconds and costs $0.034 per 1,000 images, according to TechCrunch. Google DeepMind says that is roughly five times faster than the full Nano Banana 2.
The pitch is not quality; it is throughput. Where the standard model is billed as a "generalist workhorse," Lite is tuned for jobs that need many images fast and cheap, the drafting and iteration stage rather than the finished frame. It renders only at a 1k canvas, dropping the 2k and 4k options of its larger siblings, per VentureBeat. That is the trade: resolution and polish for speed and volume.
The benchmark numbers complicate the "lite" label. In Google's internal Text-to-Image arena, the model scored an Elo of 1251, ahead of the original Nano Banana at 1151 and, narrowly, the pricier Nano Banana Pro at 1245, VentureBeat reported. Read those figures with the usual caution owed to a vendor's own leaderboard, but the direction is clear: the cheap, fast model is no longer obviously the worst one.
Availability is the tell about who this is for. Nano Banana 2 Lite went live in Google AI Studio, the Gemini API and Google's Gemini Enterprise Agent Platform, and is rolling into consumer surfaces including AI Mode in Search, the Gemini app and Photos. Google is retiring the first Nano Banana to "legacy" status and pointing developers at the new one. Alongside it, the company widened access to Gemini Omni Flash, a video model priced at $0.10 per second of output.
What Google is really selling is the cost curve. Drive the price of a generated image toward a rounding error and the question shifts from "can we make this" to "should we make thousands of these," which is precisely the anxiety behind the backlash over AI "slop" filling social feeds. Google leans commercial, marketing the tools for ads and, more contentiously, for Hollywood; its recent $75 million deal with the indie studio A24 drew real anger from fans. The economics of that fight look a lot like the compute crunch reshaping the rest of the model business, the same pressure that pushed Google to cap a rival's access to Gemini earlier this month.
"Building with generative media is often about creative iteration," Google wrote in its announcement. Iteration is the polite word. At $0.034 a thousand, the more honest one is scale, and scale is the thing the internet is still deciding whether it wanted.
Google DeepMind on Nano Banana 2 Lite
