Google makes Nano Banana 2 and Nano Banana Pro GA with video input and $0.045/$0.134 pricing

• Google's launch post moved Nano Banana 2 and Nano Banana Pro to general availability in Google AI Studio and the Gemini Enterprise Agent Platform API.
• Pricing landed in the public chatter immediately: Phil Schmid's post listed Nano Banana 2 at $0.045 per image and Nano Banana Pro at $0.134 per image.
• The real feature addition is on the Flash-tier model, where Schmid and fofrAI's demo both point to video-to-image input for context-aware image generation.
• Output sizing also got clearer at GA, because testingcatalog's summary says 1K and 2K generation are now generally available while 4K stays in preview.

You can jump from Google's announcement to the video-to-image docs, and the product split is unusually clean: the consumer-facing Gemini Omni post already showed Nano Banana-style video editing in the Gemini app, while this release turns the underlying image models into priced API surfaces.

Nano Banana 2 can now take a video file or YouTube URL as prompt context, then generate a still image, thumbnail, infographic, or comic-strip-style output from that footage, according to fofrAI's demo and Phil Schmid's pricing post.

The docs detail one practical constraint that matters more than the marketing copy: Google's video-to-image guide says long videos may need lower fps so the extracted frames stay within the context window.

Google framed the release as GA for its "best image generation models" in AI Studio and the Gemini Enterprise Agent Platform API, per the main announcement. testingcatalog's summary adds Google's standard GA promise here, enterprise-grade infrastructure and security.

The two model tiers landed with simple per-image pricing:

• Nano Banana 2, also described as Gemini 3.1 Flash Image, is $0.045 per image according to Phil Schmid's post.
• Nano Banana Pro, also described as Gemini 3 Pro Image, is $0.134 per image according to the same post.
• 1K and 2K outputs are GA for both models, while 4K output is still preview-only, per testingcatalog.

That makes this less about a new creative demo and more about a production API with published unit economics.

A day earlier, GeminiApp's Omni post pitched video style transfer in the Gemini app, and GeminiApp's availability follow-up said Omni was available globally for AI Plus, Pro, and Ultra subscribers on web and mobile.

That consumer rollout helps explain this GA move. Tulsee Doshi's thread described Omni as bringing Nano Banana editing to video, while Ethan Mollick's take argued Google has the multimodal pieces but not a single fully blended surface yet.

In other words, the app already exposed the workflow, and this week's GA announcement exposed the API, pricing, and resolution limits behind it.