What Is Nano Banana Pro? Gemini 3 Pro Image Explained (April 2026)

Nano Banana Pro is the consumer name for Google DeepMind's Gemini 3 Pro Image model, released November 20, 2025 as Google's flagship image generation and editing model. If you've seen Nano Banana Pro on Twitter, in the Gemini app, or inside Google Slides and wondered what it actually is, this guide separates what Google has officially documented from what third parties have measured.

The aim is to explain Nano Banana Pro precisely: what it is, how it differs from the regular Nano Banana line, where it runs, what it costs, and when it's worth upgrading.

TL;DR

• Nano Banana Pro is the marketing name for Gemini 3 Pro Image (model id gemini-3-pro-image-preview), released by Google DeepMind on November 20, 2025.
• Native resolution up to 4096×4096 (1K, 2K, 4K tiers), per Google DeepMind.
• Up to 5 characters consistent per workflow and up to 14 object inputs for blending.
• Third-party benchmarks show roughly 94% text-rendering accuracy (Design for Online review), a third-party measurement, not a Google guarantee.
• Available in the Gemini app, Google AI Studio, the Gemini API, Vertex AI, Workspace (Slides, Vids, NotebookLM), Google Ads, and AI Mode in Search.
• Per-image API pricing: ~$0.134 at 2K, ~$0.24 at 4K; Batch API roughly half (pricing summary).

What is Nano Banana Pro?

Nano Banana Pro is Google's public brand for the Gemini 3 Pro Image model. The "Nano Banana" name first attached to a Gemini image model that went viral on LMArena in 2025; Google adopted it officially to differentiate the consumer product from the underlying Gemini-3 image stack. The "Pro" suffix marks the larger, slower, higher-quality variant, distinct from the Flash variant Google later released as Nano Banana 2.

Three things define what Nano Banana Pro actually is, all from Google's own product pages:

1. Built on Gemini 3 Pro reasoning. Google's DeepMind page describes Nano Banana Pro as using "Gemini's state-of-the-art reasoning and real-world knowledge to visualize information." That is the difference between drawing a picture of a recipe and drawing a recipe whose ingredient list is correct.
2. Multimodal in and out. Nano Banana Pro accepts text, reference images, and structured prompts, up to 14 input images per workflow, with consistency for up to 5 characters in the output.
3. Integrated everywhere in Google. The same Nano Banana Pro model powers Slides, Vids, NotebookLM, Google Ads, and AI Mode in Search (rollout post).

When was Nano Banana Pro released?

Google DeepMind launched Nano Banana Pro on November 20, 2025 (Google blog). Independent same-day coverage from Simon Willison confirms the date.

Nano Banana Pro started in paid preview as gemini-3-pro-image-preview, available via the Gemini API, Google AI Studio, and Vertex AI on day one (developer announcement).

Key features of Nano Banana Pro, from Google's docs

The capabilities below are pulled directly from Google's own product pages, the "claimed" column in the claimed-vs-confirmed split.

Native 4K resolution

Google DeepMind documents that Nano Banana Pro can "generate crisp visuals at 1k, 2k or 4k resolution" (source). 4K is 4096×4096 native, with no upscaler. That is the single largest difference between Nano Banana Pro and the Flash-tier Nano Banana 2.

In-image text rendering

The DeepMind page positions Nano Banana Pro for "sharp, legible text" suitable for "posters, intricate diagrams, and detailed product mockups," with multilingual rendering across writing systems. Google's developer post calls out "state-of-the-art text rendering" as a headline capability (source).

Third-party measurement: Design for Online's review shows approximately 94% accuracy on in-image text rendering benchmarks, a third-party measurement, not a Google guarantee.

Multi-character and multi-object consistency

Nano Banana Pro supports up to 5 characters consistent across a workflow and up to 14 object inputs blended into a single image, per the DeepMind product page. This is the explicit campaign-spec language that makes the model a fit for series work where the same person or product needs to recur.

Web grounding via Google Search

Nano Banana Pro can connect to Google Search to pull in real-time information before generating. Google's launch blog lists examples like generating "a quick snapshot for a recipe or visualiz[ing] real-time information like weather or sports." The Gemini API exposes this as a grounding option that engages when the prompt is data-dependent.

Thinking mode

Google's developer documentation describes a "thinking" pass that runs before final generation. The model produces interim "thought" compositions internally and refines them before rendering. The effect shows up as better coherence on complex multi-element prompts (overview).

Pro camera and lighting controls

Per Google's developer post, the Nano Banana Pro API gives developers "control over the physics (lighting, camera, focus, color grading) and composition of the image" as direct prompt fields rather than buried parameters.

SynthID and C2PA watermarking

Every Nano Banana Pro image carries an invisible SynthID watermark embedded directly in the pixel data, plus C2PA metadata for provenance (developer post). Free Gemini app users see a small visible Gemini sparkle badge; that visible badge is removed for paid tiers and direct API output.

How much does Nano Banana Pro cost?

Google DeepMind's product page does not publish a Nano Banana Pro per-image figure directly. The clearest published numbers come from Google's API documentation summarized by third-party trackers:

Source: Nano Banana Pro pricing breakdown compiled from Google's developer documentation. Confirm the current rate inside Google Cloud before you build a cost model.

For consumer access, Google bundles Nano Banana Pro into the Gemini app subscription tiers. The free tier gets a small daily quota; paid plans (AI Plus, AI Pro, AI Ultra) raise both the quota and the resolution ceiling.

Where Nano Banana Pro runs

Google rolled Nano Banana Pro across its entire product surface on launch day:

• Gemini app (free and paid tiers)
• Google AI Studio at aistudio.google.com
• Gemini API: gemini-3-pro-image-preview
• Vertex AI: enterprise with provisioned throughput, pay-as-you-go, and advanced safety filters (Google Cloud announcement)
• Google Workspace: Slides, Vids, NotebookLM (Workspace updates)
• Google Ads and AI Mode in Search
• Third-party platforms: Together AI, Kie.ai, OpenRouter, PiAPI

For most production builds, the meaningful Nano Banana Pro surfaces are the Gemini API directly, Vertex AI for enterprise, or a third-party API aggregator if you're routing across providers.

Nano Banana vs Nano Banana 2 vs Nano Banana Pro

The three names tend to get confused. Here's how they actually map.

A few notes on how Nano Banana Pro relates to its siblings:

• Nano Banana Pro is not a "newer" version of Nano Banana 2. They are parallel tiers in the same generation, like Gemini 3 Pro and Gemini 3 Flash. Pro is slower, more expensive, higher quality. Flash is faster, cheaper, tuned for volume.
• Nano Banana 2 launched three months after Nano Banana Pro. Google released Pro first (Nov 2025), then the Flash-tier Nano Banana 2 in Feb 2026 with most of Pro's capabilities at roughly half the API price (TechCrunch coverage).
• Nano Banana 2 is now the default in the Gemini app (Workspace announcement). Tap the image button without picking Pro and you'll usually get Nano Banana 2.

When should you upgrade to Pro?

Use Nano Banana 2 (the Flash tier) for the everyday case: marketing creative, social posts, draft iteration, anywhere cost-per-image matters and 1K–2K is enough. Upgrade to Nano Banana Pro when one of these is true:

• You need native 4K output. Nano Banana Pro is the only one of the three that hits 4096×4096 in a single pass.
• You need explicit multi-character consistency. Nano Banana Pro publishes the up-to-5-character spec; the Flash tiers do not.
• You're rendering long, dense in-image text, like paragraphs of body copy, multilingual infographic labels, or packaging with regulatory text.
• You're using thinking mode for complex prompts, like briefs that combine a recipe, a chart, and a brand style guide in one image.
• You need pro camera controls, with explicit lighting, aperture, focal-length, and color-grading direction.

If none of those apply, Nano Banana 2 saves about half the per-image cost without much visible quality loss at 1K–2K.

How Nano Banana Pro compares to GPT Image

The other model in the same flagship tier is OpenAI's gpt-image-1.5. We've covered this in Nano Banana Pro vs GPT Image, but the short version:

• Nano Banana Pro wins on resolution. Native 4K vs a 1024×1536 cap on gpt-image-1.5.
• Nano Banana Pro wins on stated multi-character consistency. Published spec vs none.
• gpt-image-1.5 wins on draft cost. $0.009 per Low 1024×1024 image vs ~$0.13 at Pro's 1K tier.
• Both render in-image text well, far above the DALL·E 3 baseline.

The two are complements at different points in a workflow. If you're picking between them and curious about the alternative OpenAI flagship, GPT Image 2 is also worth a look.

Try Nano Banana Pro free

The fastest way to evaluate Nano Banana Pro against your own brief is to run a few prompts through it. Try Nano Banana Pro free at /nano-banana-pro. The studio runs Nano Banana Pro directly in your browser with free trial credits on signup, no Google Cloud project setup required. For the official entry points, see Google AI Studio and the Gemini app.

Frequently asked questions

What is Nano Banana Pro?

Nano Banana Pro is Google DeepMind's flagship image model, officially Gemini 3 Pro Image (gemini-3-pro-image-preview). It launched on November 20, 2025 with native 4K output, up to 5-character consistency, 14-input blending, web grounding via Google Search, and pro camera controls.

Is Nano Banana Pro the same as Gemini 3?

Not exactly. Nano Banana Pro is the image-generation variant built on Gemini 3 Pro reasoning. The Gemini 3 Pro language model and Nano Banana Pro share the same reasoning stack but are exposed as separate endpoints. One returns text, the other returns images.

How much does Nano Banana Pro cost per image?

Standard API pricing is roughly $0.134 at 2K and $0.24 at 4K, with the Batch API at about half those rates (pricing summary). For consumer access, the Gemini app's free tier includes a small daily quota; paid Gemini tiers (AI Plus at $7.99/mo, AI Pro at $19.99/mo, AI Ultra at $249.99/mo) raise both the quota and the resolution ceiling.

Where can I try Nano Banana Pro for free?

Three easy entry points: the Gemini app (free tier, small daily quota), Google AI Studio (developer playground with a free quota), and the Nano Banana Pro studio at /nano-banana-pro (free trial credits, no Google Cloud setup).

What is the difference between Nano Banana, Nano Banana 2, and Nano Banana Pro?

The original Nano Banana runs on Gemini 2.5 Flash Image (Aug 2025). Nano Banana 2 is the Flash-tier successor on Gemini 3.1 Flash Image (Feb 26, 2026). Nano Banana Pro is the Pro flagship on Gemini 3 Pro Image (Nov 20, 2025), the only one with native 4K and the fullest published consistency spec.

Does Nano Banana Pro watermark images?

Yes. Every Nano Banana Pro image carries an invisible SynthID watermark embedded by Google DeepMind, plus C2PA metadata for provenance. Free Gemini app users see a small visible Gemini sparkle badge; paid tiers and direct API usage drop the visible badge but keep the invisible SynthID.

Is Nano Banana Pro available on Vertex AI?

Yes. Google announced Nano Banana Pro enterprise availability on Vertex AI on launch day, with Provisioned Throughput, Pay As You Go pricing, and advanced safety filters.

Sources

• Nano Banana Pro: Gemini 3 Pro Image model from Google DeepMind (Google launch announcement, November 20, 2025)
• Gemini 3 Pro Image (Nano Banana Pro) (Google DeepMind product page)
• Developers can build with Nano Banana Pro (Gemini 3 Pro Image) (developer-facing launch post)
• Where to use Nano Banana Pro (Google product surface rollout)
• Nano Banana Pro available for enterprise (Vertex AI announcement)
• Workspace rollout: Slides, Vids, Gemini app, NotebookLM
• Nano Banana 2: Combining Pro capabilities with lightning-fast speed (Flash-tier successor announcement)
• Google launches Nano Banana 2 with faster image generation (TechCrunch coverage of Feb 26, 2026 release)
• Nano Banana Pro pricing breakdown (third-party pricing summary)
• Design for Online review with text-rendering benchmark
• Simon Willison: Nano Banana Pro is the best available image generation model (independent same-day review)

Last reviewed against source pages: 2026-04-18. Pricing and capability figures change periodically; confirm in the linked sources before acting on the numbers above.