Google launched Nano Banana 2 Lite on June 30, 2026, generating images in four seconds at $0.034 per image. The company also released Gemini Omni Flash, its first API-native video generation model, at $0.10 per second of output.
Key facts
- Nano Banana 2 Lite costs $0.034 per image at 1K resolution.
- Image generation takes four seconds.
- Gemini Omni Flash costs $0.10 per second of video output.
- Video clips are limited to ten seconds.
- Audio references and scene extensions are not supported yet.
Google released two new generative AI models on June 30: Nano Banana 2 Lite for fast text-to-image generation and Gemini Omni Flash for video generation and editing via the API. According to The Decoder
Nano Banana 2 Lite: Speed Over Quality
Nano Banana 2 Lite, accessible via the API as gemini-3.1-flash-lite-image, generates a 1K-resolution image in four seconds at a cost of $0.034 per image. Google positions it for "fast ideation and high-throughput developer pipelines," not for premium output. The model brings the Nano Banana family to three production tiers: Lite (speed), Nano Banana 2 (balanced), and Nano Banana Pro (quality). Google explicitly considers the original Nano Banana (Gemini 2.5 Flash Image) outdated.
The model is rolling out across Google's consumer products, including AI Mode in Google Search, the Gemini app, NotebookLM, Google Photos, Stitch, Google Flow, and Google Ads. Despite the speed focus, Google claims reliable prompt following, consistent character rendering, and readable text in generated images — unverified claims given the lack of independent benchmark scores in the announcement.
Gemini Omni Flash: First Video API from Google
First shown at Google I/O, Gemini Omni Flash is now available to developers through the Gemini API and Google AI Studio. It combines Gemini's multimodal reasoning with video generation and editing at $0.10 per second of output, matching Veo 3.1 Fast pricing. The model supports conversational video editing via natural language and can mix input formats (text, images, video).
Limitations are significant: Gemini Omni Flash currently only generates ten-second clips. Audio references and scene extensions aren't supported in the API yet. The API schema accepts video references but Google did not disclose latency or resolution details. Google recommends chaining Nano Banana 2 Lite with Gemini Omni Flash to go from a static image to an animated video.
Developer Implications
The pricing undercuts most competitors for image generation: OpenAI's DALL-E 3 costs roughly $0.04 per image at standard resolution, while Stable Diffusion 3.5 on Replicate runs about $0.008 per image but requires self-hosting for throughput. At $0.034 per image, Nano Banana 2 Lite is positioned for high-volume use cases where four-second latency is acceptable. For video, $0.10 per second matches Veo 3.1 Fast but undercuts OpenAI's Sora Turbo ($0.20 per second) and Runway Gen-3 Alpha ($0.15 per second). The ten-second clip limit, however, restricts use to short-form content — social media ads, product demos, or animated thumbnails.
The broader signal: Google is commoditizing image generation while opening the video API floodgates. The chaining recommendation suggests a pipeline play — generate cheap images, animate them into videos — which could reshape how developers build media generation workflows. Google's $11 billion per year SpaceX compute commitment (announced June 21) and its 3 million TPU booking with Intel (June 25) provide the infrastructure backbone for this push.
Unique Take
Google's strategy here is volume-first, not quality-first. By offering a $0.034 image model and a $0.10/second video model with explicit chaining, Google is betting that developer pipelines will optimize for cost and speed rather than output fidelity. This contrasts with OpenAI and Anthropic, which prioritize model quality and safety guardrails over raw throughput. The ten-second video cap is a deliberate constraint — it forces developers to treat video as a short-form additive layer, not a standalone production tool. Watch for Google to raise the clip length once it captures API share.
What to watch
Watch for Google to extend Gemini Omni Flash beyond ten-second clips, likely by Q4 2026, and for independent latency/quality benchmarks on Nano Banana 2 Lite. The chaining workflow adoption rate — how many developers actually use both models together — will signal whether Google's pipeline strategy gains traction.
Source: the-decoder.com
