DeepMind launched two new generative media models targeting both developer and consumer use cases: Nano Banana 2 Lite for high-speed, low-cost image generation, and Gemini Omni Flash for conversational video editing and creation. Both models are available immediately to developers via Google’s API platforms following the official public launch announcement.
The launches expand DeepMind’s end-to-end generative media stack for technical and non-technical users. They add faster, lower-cost options to the company’s existing Gemini Image and Veo video generation product lines.
Nano Banana 2 Lite: Low-Cost High-Speed Image Generation for Developers and Consumers
Nano Banana 2 Lite, assigned the official model ID gemini-3.1-flash-lite-image, is the fastest and lowest-cost model in DeepMind’s Gemini Image family. Per the official DeepMind announcement, it is priced at $0.034 per 1K-resolution image output and generates text-to-image results in an average of 4 seconds per request 1.
The model is positioned as the direct replacement for the legacy first-generation Nano Banana model, which uses the model ID gemini-2.5-flash-image. It offers faster generation speeds, lower per-image costs, and improved text rendering and character consistency capabilities compared to the first-generation version 1.
Immediate access to Nano Banana 2 Lite is available via Google AI Studio, the Gemini API, and the Gemini Enterprise Agent Platform for developer use cases. The model maintains strong alignment with user-provided prompts, preserves consistent character design across sequential generated images, and renders text embedded in generated visuals clearly and accurately, even with its speed and cost optimizations 1.
Alongside developer access, Nano Banana 2 Lite is rolling out to seven Google consumer surfaces: AI Mode in Search, the Gemini app, NotebookLM, Google Photos, Stitch, Google Flow, and Google Ads.
Gemini Omni Flash: Conversational Short-Form Video Editing for External Developers
Gemini Omni Flash, assigned the official model ID gemini-omni-flash-preview, is DeepMind’s first video generation and editing model available to external developers. Priced at $0.10 per second of video output, it matches the per-second cost of DeepMind’s existing Veo 3.1 Fast video model 1.
The model supports conversational video editing via natural language prompts. It accepts multimodal input references including text, image, and video to maintain scene consistency, and leverages Gemini’s built-in knowledge of history, biology, and narrative logic to generate contextually accurate 10-second video clips as of launch 1.
Current API limitations include no support for audio reference uploads, scene extension, or full processing of 3-second video reference inputs. DeepMind has stated that longer video durations are planned for future model updates 1.
Combined Workflow Value for High-Throughput Use Cases
For high-throughput use cases such as social media content pipelines or e-commerce product visualization workflows, the $0.034 per 1K-resolution image cost of Nano Banana 2 Lite reduces per-asset generation expenses for high-volume workloads. For example, an e-commerce team generating 25 1K-resolution product visuals for a holiday catalog would pay $0.85 total for image generation.
Those assets can then be edited into 10-second promotional clips via Omni Flash’s natural language interface for $1 per clip, with no specialized video editing expertise required. The dual launch lets teams prototype and iterate on both image and short-form video assets within the same Gemini ecosystem, eliminating the need to switch between separate third-party image and video generation tools.
