Gemini Diffusion vs UI-TARS-2

Gemini Diffusion

Google DeepMind

Gemini Diffusion is an experimental text and code generation model from Google DeepMind, announced at Google I/O in May 2025 as the first diffusion-based language model to achieve quality comparable to autoregressive models on standard benchmarks. Unlike transformer-based models that predict tokens sequentially left-to-right, it generates entire blocks of text by iteratively refining noise — the paradigm used in image and video generation models — enabling faster sampling speeds and stronger mid-generation error correction for code and mathematical editing tasks. At announcement it was available only as an experimental demo via waitlist, with no public API, marking it as a research milestone rather than a production deployment.

ByteDance

UI-TARS-2, released by ByteDance in September 2025, is a major generational upgrade of the UI-TARS family of GUI interaction models, with enhanced capabilities across computer control, game environments, code generation, and tool use. It targets agentic workflows requiring robust multimodal understanding of graphical interfaces across diverse application domains.