Google's Generative Media in 2026: Nano Banana 2, Veo, Lyria, and Gemini Omni
Through 2026 Google folded its image, video, and music generation into the Gemini line: Nano Banana 2 for images, Veo 3.1 Lite for video, Lyria 3 Pro for music, and Gemini Omni as an any-input, any-output model.
Across 2026 Google pulled its generative-media models into the Gemini family rather than shipping them as separate brands. Images run through Nano Banana 2, video through Veo 3.1 Lite and then the new Gemini Omni, and music through Lyria 3 Pro, with a world model, Project Genie, generating navigable scenes. The direction is a single “any input, any output” model line rather than a scatter of point tools.
What happened
The releases landed steadily. Nano Banana 2 (Gemini 3.1 Flash Image) arrived on 26 February 2026 as Google’s flagship image model, adding real-time web-grounded knowledge, multilingual text rendering, and output up to 4K. Lyria 3 Pro (25 March 2026) extended music generation to longer, structured tracks with promptable sections. Veo 3.1 Lite (8 April 2026) added a lower-cost video-with-audio model to the Veo line.
At Google I/O on 19 May 2026, Google introduced Gemini Omni, an “any input, any output” generative model that started with video and became the headline video-generation model, and expanded Project Genie to generate interactive worlds anchored to real Google Street View locations. On 30 June 2026, Gemini Omni Flash and Nano Banana 2 Lite reached the Gemini API, AI Studio, and the enterprise platform for developers, with Omni Flash priced at $0.10 per second of video.
Why it matters for builders
Two things follow. First, consolidation simplifies the stack: instead of learning separate image, video, and music APIs, you increasingly reach them through the Gemini line with one interface, which is why Gemini Omni’s “any input, any output” framing matters. Second, availability moved from consumer apps to developer APIs during the year (the June Omni Flash and Nano Banana 2 Lite release), so these are now things you can build into a product, not just try in an app.
For prototyping, you can try many of these in the browser before committing, which pairs with the free playgrounds approach. And this is one front in a broader media-generation race; see the 2026 LLM landscape for how the model families compare, and Google I/O 2026 for the rest of that event’s announcements.
Sources
- Google, “Gemini Omni Flash and Nano Banana 2 Lite” (30 June 2026): https://blog.google/innovation-and-ai/models-and-research/gemini-models/gemini-omni-flash-nano-banana-2-lite/
- Google, “Nano Banana 2” (26 February 2026): https://blog.google/innovation-and-ai/technology/ai/nano-banana-2/
- Google, “Lyria 3 Pro” (25 March 2026): https://blog.google/innovation-and-ai/technology/ai/lyria-3-pro/
- Google DeepMind, Veo 3.1 Lite model card (8 April 2026): https://deepmind.google/models/model-cards/veo-3-1-lite/
- Google DeepMind, “Project Genie expands” (19 May 2026): https://blog.google/innovation-and-ai/models-and-research/google-deepmind/project-genie-expands/
Further reading
- Google I/O 2026 : the wider set of announcements, including Gemini Omni’s debut.
- Free AI and cloud playgrounds : where to try generative media before you build.
- The 2026 LLM landscape : how the model families compare.