On June 30, Google published a blog post introducing Nano Banana 2 Lite, model name gemini-3.1-flash-lite-image. It is positioned as the fastest, most cost-efficient image model in the Nano Banana family: text-to-image output in roughly four seconds, at $0.034 per 1K-resolution image, aimed at high-throughput, rapid-iteration workloads. The post is credited to Alisa Fortin and Anish Nangia, product managers at Google DeepMind, and it also announced that the Gemini Omni Flash video generation model is entering public preview at $0.10 per second of output — the same price as Veo 3.1 Fast.
The real story is not a single model but a standardized workflow: generate fast stills with Lite, then set them in motion with Omni. Google’s own framing is blunt — “the real magic happens when you chain these models together.”
The Economics of Four-Second Images
The $0.034 figure changes behavior. When a draft image costs about a cent, “generate ten and pick one” stops being a luxury and becomes the default loop for interactive prototyping. At batch scale the arithmetic compounds quickly: a thousand drafts cost $34, which puts aggressive exploration inside almost any content or product team’s budget. Google stresses that Lite keeps prompt adherence, character consistency, and legible in-image text despite the speed focus — the three capabilities that fast-draft pipelines usually sacrifice first.
Google also marked the original Nano Banana (Gemini 2.5 Flash Image) as a legacy model and explicitly recommends migrating to Lite. Two years ago that was flagship image capability; today it is the entry-level replacement. There is no cleaner footnote to how fast the cost curve in image generation has fallen.
Family Division of Labor: Lite, 2, and Pro
Google now splits image generation into three tiers: Nano Banana 2 Lite for speed and scale, Nano Banana 2 as the balanced “workhorse” for everyday generation, and Nano Banana Pro holding the quality ceiling for complex professional work. When Nano Banana 2 launched in February, the pitch was “Pro-level capability, Flash-level speed.” Lite now pushes the cost frontier another step down, and the family’s price bands and use cases are cleanly separated.
For developers the practical effect is straightforward: drafts, bulk generation, and final output can route to different models, with one API managing three cost structures.
Gemini Omni Flash and Model Chaining
Omni Flash (gemini-omni-flash-preview) is a video generation and conversational editing model, first shown at Google I/O. It supports multimodal referencing — text, image, or video as input — carries Gemini’s real-world knowledge, and can sync text and graphics to actions in the video. Conversational editing means you revise the output in natural language instead of re-prompting from scratch, which is what makes the stills-to-video chain feel like one continuous workflow. The limits matter just as much: generations are currently capped at ten seconds, audio reference uploads and scene extension are not supported, video references are not yet processed correctly, and character consistency can falter across scene changes and camera pans.
The chaining takes concrete shape in three remixable demo apps. Omni Product Studio turns still product images made with Lite into cinematic e-commerce videos; the other two are Anywhere and Space Lift. An Interactions API supports up to three sequential edits per session, so image, edit, and animate can happen inside a single flow.
Rollout, Watermarking, and the Slop Debate
Lite is available now in Google AI Studio, the Gemini API, and the Gemini Enterprise Agent Platform, and is rolling out across consumer surfaces including AI Mode in Search, the Gemini app, NotebookLM, Google Photos, Stitch, Flow, and Google Ads. Omni Flash is in public preview in AI Studio, the API, and the Agent Platform, plus the Gemini app and Flow. The pattern is familiar from every previous Gemini model generation: developers and enterprise users get access first through the API surface, and consumer products absorb the model over the following weeks.
Both models carry SynthID watermarking, verifiable through the Gemini app, Chrome, and Search. The timing deserves context: reporting cited by TechCrunch estimates that about 60 percent of TikTok videos are already AI-generated, and Google’s $75 million partnership with A24 drew fan backlash. When generation gets an order of magnitude cheaper, whether “ten times cheaper” capability becomes a slop factory is the long-term question for this product line.
Sources
- Gemini Omni Flash and Nano Banana 2 Lite — Google
- Google introduces a faster, cheaper image generator with Nano Banana 2 Lite — TechCrunch
- Google launches Nano Banana 2 — TechCrunch
AI-assisted summary compiled from the sources above, reviewed by a human before publishing.