Gemini Diffusion vs GLM 5

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.

Zhipu AI

GLM-5 is a large language model from Zhipu AI with 745 billion total parameters and 44 billion active parameters per inference, designed for agentic intelligence, advanced reasoning, and frontier-level performance. It features a 200K token context window and targets complex multi-step tasks requiring sustained reasoning and tool use. GLM-5 is available via Zhipu AI's Z.ai platform under an MIT license.