Google Labs on October 7 launched Playground, a browser-based platform that lets users build playable games from text prompts without writing code1,2. The tool is available to US users aged 18 and older and is powered by three Google models: Gemini, Nano Banana, and Lyria.
Creators choose a genre, such as trivia or racing, or start from scratch, then specify whether the game should be 2D or 3D and single-player or multiplayer. They describe the desired mechanics, gameplay, and visual style in natural language; Playground also accepts uploaded visuals, which its AI transforms into game assets matched to the project's art direction.
Finished games run on both mobile devices and computers. Projects can stay private, be shared via link, or be published to Playground's Explore gallery, where they can support features such as leaderboards. Creators can continue modifying published games after release.
Google indicated that Unity Spark integration is planned for a future update.
ANALYSIS Playground extends Google's pattern of surfacing its generative-AI stack through consumer-facing tools, this time bundling a large language model (Gemini), a less-documented model (Nano Banana), and its music-generation model (Lyria) into a single creative workflow. The addition of Unity Spark integration would connect the platform to an established game-engine ecosystem, broadening the range of outputs beyond browser-native formats.