DreamCore AI Game Platform Adds Google Gemini 3.5 Flash Model
The integration signals a shift toward faster, real-time game creation from natural language alone, with DreamCore adopting a hybrid model strategy to avoid vendor lock-in.
Reporting from 1 sources: GameBusiness.jp.
NEIGHBOR announced on May 22, 2026, that its AI game generation platform DreamCore has integrated Google's Gemini 3.5 Flash model. The update aims to increase generation speed, handle complex game logic, improve mobile optimization, and strengthen multilingual support. DreamCore generates code, images, 3D models, and sound from text or voice commands.
NEIGHBOR integrated Google's Gemini 3.5 Flash into DreamCore on May 22, 2026, replacing the platform's core reasoning model. The company expects the switch to make the 'dialogue to generation to play' loop closer to real-time, with faster game template generation. DreamCore's proprietary AI agent controls code, images, 3D models, and sound generation from a single text or voice prompt, then lets users play and share the result in a browser.
The platform has seen real-world use: TV Tokyo used it for program promotion games that drew 17 million impressions on X. NEIGHBOR said it plans to keep DreamCore model-agnostic, switching or running multiple models as newer foundation models appear. The Gemini 3.5 Flash update also improves handling of complex game logic like multi-character interactions and stage progression, and generates mobile-first UIs with vertical screens and touch controls.
Synthesized by Yomimono from the 1 cited source below, including Japanese-language reporting where cited, then editorially reviewed before publishing.