Gemini Omni can take photos, video and audio and build entirely new scenes.

Judge as of 2026-05-19 against what Google/Google DeepMind documented for the model called Gemini Omni at or around that date. The claim has two parts: (a) the model accepts photos, video and audio as inputs; (b) it can generate entirely new scenes (new image or video content, not just editing or describing the inputs). TRUE: official documentation, a launch post, a model card or credible hands-on reporting shows a product named Gemini Omni that existed at that date, accepted all three input types and generated new scenes. MOSTLY TRUE: both parts hold, with a minor slip, such as one input type supported only in a limited form (for example, audio only through a video's soundtrack) or the capability limited to a preview tier. MIXED: scene generation is confirmed, but one of the three input types is clearly not supported, or the inputs are confirmed but generation is only partial (for example, editing existing scenes rather than building new ones). MOSTLY FALSE: the product exists but does only a small part of what is claimed, for example it understands those inputs but does not generate new scenes. FALSE: no product called Gemini Omni existed at that date, or official documentation shows it lacks the claimed capabilities.

Set before the research began.

1. E1 Introducing Gemini Omni Google · 19 May 2026 Own record

   “With Omni, you can combine images, audio, video and text as input and generate high-quality videos grounded in Gemini's real-world knowledge.”

2. E2 Introducing Gemini Omni Google · 19 May 2026 Own record

   “While only voice references will be supported for audio to start, we’ll roll out other types of audio inputs soon.”

3. E3 Introducing Gemini Omni Google · 19 May 2026 Own record

   “In time we will support output modalities like image and audio.”

4. E4 Gemini Omni 1.1 Flash Preview | Gemini Enterprise Agent Platform | Google Cloud Documentation Google Cloud · 27 August 2026

   “Gemini Omni 1.1 Flash (Preview) is a multimodal model designed for video, image, and text tasks.”

Google's dated launch post for Gemini Omni (E1) says that on 2026-05-19 the model could take images, audio, video and text as input and generate new high-quality videos. That satisfies both the input part and the scene-generation part of the claim. However, the same post says that at launch audio input was limited to voice references, with other audio types to come later (E2). The criteria treat an input type supported only in a limited form as a minor slip, which makes this mostly true. E1 shows video as the generated output at launch, and E3 says image and audio output will be supported in time.