Google: Nano Banana 2 and Nano Banana Pro go GA with video-to-image generation
On May 28, 2026, Google declared general availability (GA) of two native visual models: gemini-3.1-flash-image (Nano Banana 2) and gemini-3-pro-image (Nano Banana Pro). The novelty is video-to-image generation in the flash variant, which generates thumbnails, movie posters and infographics from video files or public YouTube URLs.
This article was generated using artificial intelligence from primary sources.
On May 28, 2026, Google announced the general availability (GA — General Availability) of two native visual models from the Gemini API ecosystem. The transition from preview to production phase means teams can build commercial applications relying on the stability of the API.
Which models go GA?
gemini-3.1-flash-image, code-named Nano Banana 2, is an efficient option for image generation — lower latency and lower compute cost make it suitable for high-volume applications. gemini-3-pro-image, code-named Nano Banana Pro, targets high performance and output quality for more demanding creative applications.
Both models are available to development teams through the standard Gemini API, which allows integration into existing multimodal pipelines that already use Gemini models for text or audio.
How does Nano Banana 2 generate images from video?
The most significant new capability of the flash variant is video-to-image generation. Developers can pass a video file as a multimodal input — via direct upload or as a public YouTube URL — alongside a text prompt, and the model generates visual derivatives from that content. According to the release notes, the feature is supported exclusively on the gemini-3.1-flash-image model.
Google cites three applications: generating high-quality thumbnails, creating cinematic movie posters and producing concise infographics from video content. This extends the Gemini API beyond the static text-to-image approach toward dynamic video as a source.
What happens to the preview versions?
The preview versions gemini-3.1-flash-image-preview and gemini-3-pro-image-preview have been declared deprecated with a shutdown date of June 25, 2026. Developers who use the preview model IDs in production code must migrate to the GA versions before that date to avoid service interruption. The GA versions bring stable API contracts and long-term support, unlike the preview versions, which are subject to change without notice.
Frequently Asked Questions
- What are Nano Banana 2 and Nano Banana Pro?
- They are the code names of two of Google's native visual models from the Gemini API: gemini-3.1-flash-image (Nano Banana 2) for efficient image generation and gemini-3-pro-image (Nano Banana Pro) for high performance. Both reached general availability (GA) on May 28, 2026.
- How does video-to-image generation work?
- Users can pass a video file via direct upload or as a public YouTube URL as additional context alongside a text prompt. The feature is supported exclusively on the gemini-3.1-flash-image model and generates thumbnails, movie posters and concise infographics.
- When are the preview versions of the models shut down?
- The preview versions gemini-3.1-flash-image-preview and gemini-3-pro-image-preview have been declared deprecated with a shutdown date of June 25, 2026. Developers must migrate to the GA versions before that date.
Sources
📬 AI news in your inbox
A daily digest built your way — pick topics, sources and cadence. One-click unsubscribe.