Google continues to develop Sparkify experiments, first revealed in Google I/O and is only available on WaitList. Sparkify is positioned as a standalone tool, rather than just an extension of previous projects like Google Illuminate. The interface provides three main sections: Discover, Generate, and Library.
The Explore page usually features a short video with a roughly 2-minute themed, including the origins of drinks and quirky landmark tours, all created with AI, covering a variety of visual styles, including manga, origami, anime and realistic. The video can be vertical or horizontal, and is very similar to the format found on Tiktok or Instagram reels. Each video combines sound and visual to tell a concise story, explains topics, and consistent use of specific characters across multiple videos suggests a rapid, drive or template-based approach for character persistence.
@testingcatalog A leaked video sample from a closed Google Labs experiment shines. Episode 2 #AI #VEO #GOOGLE
♬Original Sound-TestingCatalog
Sparkify's core relies on Google's latest AI models, Gemini and Veo. Including Gemini is suggested by a more refined pipeline and could handle content assembly or narrative structures, but VEO promotes real video generation. When you try out the Generation feature, users will be presented with controls that set topics, visual styles, and moods. Options include felt, clay, origami, 3D manga, anime, and moods such as realistic, fun, academic, funny and more. However, at this stage, the actual output is a static demo video, not yet responding to user prompts, leaving key questions about quick validity and model constraints unanswered.
Given the length and quality of Sparkify's output, there is speculation that it will generate shorter clips and sew them where the Gemini orchestration will likely enter. This approach allows you to create longer and more cohesive stories while remaining within the limits of current generation AI for video durations. Character design consistency throughout the video supports the idea that creators can develop social media brand personalities and themes and give power to new types of AI-driven content channels.
Breaking🚨: An early look at Sparkify, a waitlist experiment at Google Labs.
Sparkify uses a combination of VEO 3 and Gemini models to create a stunning 2-minute information video on selected topics. pic.twitter.com/czqhlpndrz
– TestingCatalog News 🗞 (@TestingCatalog) June 7, 2025
For now, Sparkify remains an invitation-only experiment without a timeline for public releases. The current status reflects Google's continuous focus on AI-driven media tools, providing a glimpse into how Gemini and VEO are leveraged to enable rapid, thematic video creation for education and entertainment. TestingCatalog is monitoring the capabilities of video generated by these AI on platforms like Tiktok as Google continues to fine-tune the experience before the wider rollout.
