Nano Banana 2 Lite and Omni Flash: 4-Second Images, $0.10/Video Second

Four seconds and $0.034 per image—Nano Banana 2 Lite's numbers make it the obvious drop-in replacement for anyone still on the legacy Nano Banana model. Google DeepMind released the model (gemini-3.1-flash-lite-image) today alongside Gemini Omni Flash (gemini-omni-flash-preview), a video generation and editing model at $0.10 per second of output.

What Nano Banana 2 Lite Actually Delivers

Nano Banana 2 Lite targets high-throughput pipelines where latency and cost are the primary constraints. At 4 seconds for text-to-image generation and $0.034 per 1K-resolution image, it's significantly faster and cheaper than its predecessor. Benchmarks show it retains reliable prompt adherence, character consistency, and legible in-image text—trade-offs I'd expect from a speed-optimized model. The recommended upgrade path is for developers currently using gemini-2.5-flash-image (the original Nano Banana) to swap to this new version for immediate gains in quality, speed, and cost.

Two other models fill out the family: Nano Banana 2 (gemini-3.1-flash-image) as the generalist workhorse, and Nano Banana Pro (gemini-3-pro-image) for complex professional tasks. But for rapid ideation and high-volume drafting, Lite is the one to reach for.

Omni Flash: Video Generation as a Conversational Tool

Gemini Omni Flash brings multimodal video generation and editing to the Gemini API and Google AI Studio for the first time. At $0.10 per second of video output—same as Veo 3.1 Fast—it supports conversational editing: describe changes in natural language, and the model modifies the video accordingly. It combines text, image, and video inputs, leveraging Gemini's world knowledge (history, biology, narrative logic) to construct coherent scenes. Text and action synchronization is supported through simple prompting.

Current limitations: 10-second maximum video duration, no audio reference uploads, and video references up to 3 seconds aren't correctly processed yet. Google promises longer durations soon. For many use cases—short social clips, rapid prototypes, iterative storyboarding—this is already functional.

Where These Models Fit in the Pipeline

Both models are available today in Google AI Studio, the Gemini API, and the Gemini Enterprise Agent Platform. Consumer surfaces like AI Mode in Search, Gemini app, NotebookLM, and Google Photos are also getting Nano Banana 2 Lite. The real play is chaining them: generate a batch of 1,000 images for $34, then feed select images into Omni Flash to create short video sequences with conversational edits. One API, two models, aggressive pricing—Google DeepMind is betting on volume over margin. I'd start experimenting today to see how well the latency and quality hold up under real workloads.

Source: Start building with Nano Banana 2 Lite and Gemini Omni Flash
Domain: deepmind.google