Today Google released two generative-media models for developers: Nano Banana 2 Lite, a fast, cost-efficient image model, and Gemini Omni Flash, a multimodal model for high-quality video generation and conversational editing. Both models are available via Google AI Studio, the Gemini API and the Gemini Enterprise Agent Platform, and are being integrated into several Google consumer products.
Nano Banana 2 Lite — fast, cost-efficient image generation
Nano Banana 2 Lite (gemini-3.1-flash-lite-image) is the fastest and most cost-efficient image model in the Nano Banana family to date, designed for high-throughput, low-latency workflows. Google recommends developers currently using the original Nano Banana (gemini-2.5-flash-image) to upgrade.
Key performance figures (as reported by Google):
- Latency: text-to-image outputs in about 4 seconds.
- Cost-efficiency: $0.034 per 1K-resolution image.
- Quality: despite prioritizing speed, the model maintains strong prompt adherence, character consistency and readable in-image text rendering.
Overview of the Nano Banana family according to Google:
- Nano Banana 2 Lite (Gemini 3.1 Flash Lite Image): optimized for speed and near-real-time, high-volume pipelines.
- Nano Banana 2 (Gemini 3.1 Flash Image): a generalist workhorse balancing quality and latency.
- Nano Banana Pro (Gemini 3 Pro Image): targeted at complex professional use cases requiring accuracy and control.
- Nano Banana (Gemini 2.5 Flash Image): legacy model; Google recommends upgrading.
Nano Banana 2 Lite will also appear on consumer surfaces such as AI Mode in Search, the Gemini app, NotebookLM, Google Photos, Stitch, Google Flow and Google Ads.
Gemini Omni Flash — multimodal video generation and conversational editing
Gemini Omni Flash (gemini-omni-flash-preview) brings Gemini’s multimodal reasoning to video generation and editing. The model is now available to developers in Google AI Studio, the Gemini API and the Gemini Enterprise Agent Platform; it is also available in the Gemini app and Google Flow.
Pricing and capabilities:
- Pricing: $0.10 per second of video output (Google notes this matches the price of Veo 3.1 Fast).
- Capabilities: conversational video editing via natural language, multimodal referencing (combining text, image and video inputs), incorporation of real-world knowledge (e.g., history, biology, narrative logic), and text-action synchronization for connecting prompts to visual actions.
Limitations (as reported by Google):
- Current video generation is limited to 10-second clips; longer durations will be added later.
- Uploading audio references and scene extension are not yet supported in the Gemini API for this model.
- The API schema accepts video references up to 3 seconds, but those are not currently processed correctly by the model.
- Character consistency across scene changes or panning movements has limitations that Google is addressing.
Google DeepMind has published detailed benchmarking information for Gemini Omni.
Chaining the models and demo apps
Google highlights that combining the models enables end-to-end multimedia workflows: use Nano Banana 2 Lite for rapid image generation and pass those images as references to Gemini Omni Flash to animate them into video. The Interactions API supports multi-turn experiences with preserved session history and context, allowing up to three sequential edits.
Demo applications provided by Google include:
- Anywhere: a demo that transforms a selfie or uploaded photo into dozens of iconic locations using Nano Banana 2 Lite, then uses Omni Flash to turn a selected generated image into an animated clip.
- Space Lift: an interior-design demo that generates multiple design concepts from a photo with Nano Banana 2 Lite and then produces a cinematic video showcase using Omni.
- Omni product studio: converts static images made with Nano Banana 2 Lite into cinematic e-commerce videos via Gemini Omni.
Safety, transparency and verification
Both models run on Google’s secure infrastructure and use SynthID watermarking to mark AI-generated content. Verification of AI-created content is available in the Gemini app, Gemini in Chrome and in Search. Google says it is expanding verification tools to make it easier to understand how content was created and edited across the web.
Getting started
Google recommends developers try the models in Google AI Studio’s playground and consult the Gemini API documentation and the respective prompting guides for Nano Banana 2 Lite and Gemini Omni Flash. Example code and remixable demo apps are provided to help developers combine the models into fast, iterative multimedia workflows.
