Google Launches Nano Banana 2 Lite at $0.034 per 1,000 Images

Google launched Nano Banana 2 Lite on Tuesday, a faster and cheaper image generation model priced at $0.034 per 1,000 images. The model, designated gemini-3.1-flash-lite-image, can produce text-to-image outputs in four seconds, the company said in Nano Banana 2 Lite’s pricing, speed, and lineup details. It is rolling out across Google AI Studio, the Gemini API and the Gemini Enterprise Agent Platform, with consumer-side expansions in AI Mode in Search, the Gemini app and several other Google products.

Google built the Lite tier for the same work its bigger siblings already do at higher quality: rapid creative iteration on commercial workloads. In the company’s framing, the model is what you pick when speed and budget matter more than maximum fidelity. The launch targets A/B testing of ad variations, social platforms serving many users, and product teams that need to draft at pace. The move also pushes the original Nano Banana, now labelled a legacy model, out of the lineup, and points its users at the Lite tier as a direct upgrade.

The Price and Speed That Frame the Story

Both numbers carry the launch pitch. At $0.034 per 1,000 images, a campaign that runs 10,000 visual variants costs about 34 cents to render, before any retries or A/B splits. The four-second turnaround is what Google is betting will let creative teams sketch inside the same loop they write copy. The model keeps reliable prompt adherence, strong character consistency and legible in-image text, the basics a creative team needs to evaluate work in progress.

Google positions the Lite tier as a high-velocity drafting tool rather than a destination renderer. It isn’t pitched at hero shots, where Nano Banana 2 and Nano Banana Pro still sit. The trade-off is explicit: cheaper and faster at the front end, less fidelity at the back.

The same launch debuts Gemini Omni Flash in public preview at $0.10 per second of video output, a price Google says matches Veo 3.1 Fast. Running image and video generation through one billing layer is the quieter part of the rollout, and the part most enterprise customers will notice second.

Where It Sits in the Nano Banana Family

The Nano Banana line now has four members, and Google splits them by how much speed matters relative to polish. Nano Banana 2 Lite is the new entry point for high-volume workflows. Nano Banana 2 sits above it as a generalist workhorse for a balance of quality and cost. Nano Banana Pro takes the top tier for professional use cases that demand accuracy and reasoning. The original Nano Banana, now formally Gemini 2.5 Flash Image, becomes the legacy model Google is asking existing customers to leave behind.

Google’s recommendation for current Nano Banana users doesn’t fix one tier. It points to Lite for most. The roster reads in order of latency, with each step up adding capability and cost:

1. Nano Banana (Gemini 2.5 Flash Image): legacy model, recommended replacement is Nano Banana 2 Lite
2. Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image): high-volume tier, four-second generation, $0.034 per 1K image
3. Nano Banana 2 (Gemini 3.1 Flash Image): generalist workhorse, balanced quality and cost
4. Nano Banana Pro (Gemini 3 Pro Image): professional tier for complex compositions and text-heavy layouts

For adopters, the swap from the original Nano Banana to Nano Banana 2 Lite is positioned as immediate across key performance dimensions, according to the company’s blog. Google hasn’t given a fixed sunset date for the legacy tier, but the legacy label carries that signal.

The Cost Race Against Other Image APIs

At $0.034 per 1,000 images, Nano Banana 2 Lite sits at the bottom of Google’s own list. Vertex AI’s older Imagen 4 line still lists its Fast tier at $0.02 per image, Standard at $0.04, and Ultra at $0.06, all per 1024-by-1024 image. OpenAI’s GPT Image 1 ranges from $0.011 at low quality to $0.167 at high quality for a 1024-by-1024 image, with a Mini variant spanning $0.005 to $0.036. Per-image sticker prices don’t all measure the same thing.

Per-image rates don’t map cleanly onto the same output at every provider. Google is selling the Lite tier against throughput, since the four-second generation time is what makes the volume math work for marketing pipelines. The same draft-and-discard workflow that costs $0.034 per 1,000 on Nano Banana 2 Lite can quickly become the more expensive option on an Imagen or GPT Image plan when a campaign runs many low-quality throws. The 2026 image-generation price competition is increasingly about throughput rather than per-image sticker price, according to a 2026 Google vs OpenAI image pricing analysis.

Who Is Already Plugging It In

The launch lands with a roster of brand-name adopters already committed to testing or shipping. Google published quotes from Adobe, Figma, Artlist and Manus AI alongside the rollout. WPP and Invideo weighed in on the companion video model, Gemini Omni Flash. Each frames Nano Banana 2 Lite as a tool to compress iteration cycles inside an existing workflow, not a destination product a user opens on its own.

Speed is no longer a limitation. When generation is faster than imagination, creators can stay inside the idea instead of waiting on the tool.

Idan Yonas, director of AI content and innovation at Artlist, made the comment in Google’s Nano Banana 2 Lite pricing and enterprise rollout details post. His team was working through how the four-second turnaround reshapes the loop between prompt and review, where seconds saved per draft change how many variants a creative team is willing to try.

Figma co-founder and creative director Itay Schiff said the model helps designers stay in the creative flow on Figma Weave’s node-based canvas. Manus AI co-founder and chief product officer Tao Zhang said his team tested it for slide decks and web pages in autonomous agent workflows, where the seconds saved per iteration compound across runs. Adobe senior director of product Matt Chotin said Firefly would integrate Nano Banana 2 Lite alongside Gemini Omni Flash, pairing Adobe’s pro-grade tools with leading creative AI models.

The validation comes from partners Google itself picked to announce with, and the quotes Google chose to print. Each description treats the Lite tier as something to fold into an existing tool rather than open on its own. The named companies are platforms that aggregate creative demand at scale: Figma for design, Manus AI for autonomous workflows, Artlist for stock and creator tooling, Adobe for the Firefly creative suite.

The Backdrop Google Is Choosing Not to Lead With

The launch lands inside a broader fight over AI-generated imagery. Critics use the term AI slop to describe the volume of low-quality synthetic content now circulating online. Advertisers and marketing teams continue to invest in image tools for the operational benefits of rapid content testing, but the public conversation around those tools has grown louder and harsher. Google often markets its models as convenience tools that assist ad creation rather than replace it, an effort to keep the conversation on productivity rather than on volume itself.

On the same week as the launch, public attention also fell on a $75 million partnership between Google and indie studio A24, a deal that has drawn significant criticism from fans and creative communities. The A24 collaboration is separate from the Nano Banana family and pitches Google deeper into entertainment at a moment when its image generation tools are feeding more pixels into the wider web. The pattern is now familiar: Google’s image generation pitch, its entertainment investments and its ethics critics all sit in the same news cycle. A similar tension surfaced earlier this year when Google Android security chief resigned over Pentagon AI, citing a lost moral compass.

Google’s counter is procedural. Both Nano Banana 2 Lite and Gemini Omni Flash ship with C2PA content credentials and imperceptible SynthID watermarks turned on by default, so downstream tools can verify content origin. The protections do not address the criticism of oversaturation directly. They give enterprise customers and regulators an audit trail for the moment a generated asset becomes a question of provenance.

What Users of the Original Nano Banana Should Know

Google hasn’t given a specific date for ending support of the original Nano Banana, but the legacy label signals one is coming. Existing users get three replacement choices, and Google’s own recommendation points to the Lite tier for most. Migration is a swap on the developer side, since the model’s API endpoints follow the same authentication and billing setup as the rest of the Gemini lineup.

The right tier depends on the workflow:

• Pick Nano Banana 2 Lite for rapid ideation, A/B testing ad variations, and high-volume social pipelines, where Google’s positioning puts it
• Pick Nano Banana 2 when you want the generalist balance the Lite tier trades away for speed and cost
• Pick Nano Banana Pro for text-heavy layouts, complex compositions, or any task where accuracy matters more than seconds per render

The enterprise layer is where the headline numbers land first. The Gemini Enterprise Agent Platform offers provisioned throughput for Nano Banana 2 Lite from launch day, designed for high-concurrency API requests at scale. Developers can begin with developer docs for the Gemini image API to swap from the legacy model to the new tier, with usage and billing managed through the Google Cloud console.

Frequently Asked Questions

What is Nano Banana 2 Lite?

Nano Banana 2 Lite is Google’s newest entry in the Nano Banana image generation family and the second model built on the Gemini 3.1 Flash line. Google describes it as the fastest and most cost-efficient image model in the family, designed for high-volume workflows where speed and cost matter more than maximum fidelity, with text-to-image outputs in four seconds at $0.034 per 1,000 images.

How much does it actually cost?

Google’s launch post prices Nano Banana 2 Lite at about $0.034 per 1,000 images at 1K resolution. The Gemini API pricing page lists the same tier at $0.0336 per 1K-resolution image, the small gap explained by Google’s about framing on the blog figure. For very high concurrency, provisioned throughput is available on the Gemini Enterprise Agent Platform from launch day.

Is Google shutting down the original Nano Banana?

Google has labelled the original Nano Banana, formally Gemini 2.5 Flash Image, as a legacy model and recommends migrating to Nano Banana 2 Lite, Nano Banana 2 or Nano Banana Pro. The company hasn’t announced a specific end-of-support date, but the legacy designation is a clear signal that one is coming.

What can I do with Nano Banana 2 Lite that I could not with the original?

Google positions the Lite tier on three advantages over the original: faster generation, lower per-image cost, and stronger character consistency with legible in-image text. The model targets high-velocity pipelines such as A/B testing of ad variations, social platforms serving many users, and rapid storyboarding, where iteration speed rather than maximum fidelity is the binding constraint.

Where do I access it?

Three channels carry the model at launch: Google AI Studio for prompt testing, the Gemini API for developer integration, and the Gemini Enterprise Agent Platform for business workflows. Google is also rolling it out across consumer surfaces including AI Mode in Search, the Gemini app, NotebookLM, Google Photos, Stitch, Google Flow and Google Ads.