Nano Banana v2

Overview

Nano Banana 2 (officially Gemini 3.1 Flash Image) is Google DeepMind's latest state-of-the-art image generation model, released on February 26, 2026. It bridges the gap between the original Nano Banana Pro's advanced intelligence and the rapid iteration speed of Gemini Flash, making high-quality image creation accessible to every Gemini user — not just paying subscribers.

The model introduces real-time web-search grounding, subject consistency across complex multi-character scenes, production-ready resolution controls up to 4K, and significantly improved visual fidelity. Nano Banana 2 is now the default image model across the Gemini app, Google Search, and Flow, replacing Nano Banana Pro as the standard offering for most users. It also powers creative suggestions during campaign creation in Google Ads.

What's New

Flash-Speed Intelligence with World Knowledge Grounding

Nano Banana 2 brings Gemini Flash's speed together with search-grounded generation — making real-time web grounding available at Flash-tier latency. When generating specific subjects — landmarks, products, public figures, or branded assets — the model is informed by real-time text and image search results, producing more accurate and detailed results than earlier Nano Banana versions. This grounding also enables new use cases like generating data visualizations, infographics, and note-to-diagram conversions that require factual accuracy.

Subject Consistency at Scale

One of the most significant upgrades over the original Nano Banana is subject consistency. Nano Banana 2 can maintain the visual identity of up to 5 characters and the fidelity of up to 14 objects across a single multi-image workflow. This makes it practical for storyboarding, sequential content, character sheets, and multi-angle product catalogs where visual continuity is critical. Previous Nano Banana versions struggled to preserve character appearance across more than 2–3 generations.

Precise Instruction Following

Nano Banana 2 demonstrates significantly improved adherence to complex, multi-layered prompts. The model captures nuanced compositional details — specific angles, lighting conditions, spatial relationships, stylistic references — with fewer iterations required. Users producing mockups, advertising visuals, or editorial content will find less prompt engineering necessary compared to Nano Banana (1.0.0) or competing AI image generators.

Production-Ready Specs: 512px to 4K

Full control over output resolution and aspect ratio is now available, ranging from 512px (lightweight for digital use) to 4K (print-ready). Users can select from a wide range of supported aspect ratios — including 1:1, 16:9, 9:16, 4:1, 1:4, 8:1, and 1:8 — enabling assets to be reused across social, digital, and print channels without re-generation. This brings Nano Banana 2 in line with tools like FLUX.2 and Ideogram 3.0 in terms of professional output specification.

Precision Text Rendering and Translation

Nano Banana 2 inherits Nano Banana Pro's text rendering capabilities at Flash speed. It generates legible, accurately placed text for posters, greeting cards, UI mockups, and marketing materials. Crucially, the model can localize text within an image — translating copy into a target language while preserving layout, style, and visual context — useful for multilingual campaigns without requiring additional design tools.

Visual Fidelity Upgrade

Beyond speed improvements, Nano Banana 2 delivers vibrant lighting, richer textures, and sharper details compared to the original Nano Banana. Photorealistic outputs in particular show marked improvement, with more natural skin tones, accurate material reflections, and better handling of environmental lighting. The upgrade reduces the visual quality gap previously observed between Nano Banana and Nano Banana Pro.

Pricing & Plans

Nano Banana 2 is available through multiple access points, with pricing depending on the platform.

• Gemini App (Free tier) — Nano Banana 2 is the new default image model across Fast, Thinking, and Pro model modes at no additional cost. Availability and generation quotas vary by region and plan.

• Google AI Plus / AI Pro / AI Ultra — Tiered monthly plans (see Gemini subscriptions for current pricing; promotional rates may apply). Subscribers retain access to Nano Banana Pro for high-fidelity specialized tasks by using the three-dot regeneration menu in the Gemini app. Nano Banana 2 serves as the default for standard workflows.

• Gemini API / AI Studio (Preview) — Pay-per-use, billed at $60 per 1M output tokens. Output cost varies by resolution at Standard rates: ≈$0.045 (512px), ≈$0.067 (1K), ≈$0.101 (2K), ≈$0.151 (4K); Batch pricing is lower. Image Search Grounding incurs an additional charge. See the Gemini API pricing table for details. Also accessible via Google Antigravity.

• Google Cloud / Vertex AI (Preview) — Enterprise pricing through Vertex AI. Contact Google Cloud sales for volume and committed-use discounts.

• Flow — Nano Banana 2 is the default image model in Google's Flow platform, available to all Flow users at zero credits.

• Google Ads — Integrated directly for campaign creative suggestions; no separate image generation billing for Ads usage.

Best For

• Content creators publishing 10+ visuals weekly across social media, YouTube thumbnails, and editorial images who need consistent quality without long wait times.
• Marketing teams prototyping multilingual campaign assets — posters, ad creatives, localized visuals — without dedicated design tools or translation workflows.
• Product teams and e-commerce operators maintaining visual consistency across product catalogs, multi-angle shots, and lifestyle scene compositions.
• Storytellers and storyboard artists requiring character continuity across multi-scene workflows without manually fixing visual drift between frames.
• Developers integrating image generation into apps via the Gemini API who need a high-quality, cost-efficient Flash-tier model with production-ready specs.
• Google Ads advertisers seeking automated, on-brand creative suggestions during campaign setup without additional tooling.

FAQ