DiffusionGemma

Google has released the open-weight DiffusionGemma model, specifically the 26 billion parameter variant (google/diffusiongemma-26B-A4B-it), under the Apache 2 license. This model is hosted on NVIDIA's NIM cloud API, demonstrating performance of at least 500 tokens/second in text generation tasks, which is a significant improvement over the previous Gemini Diffusion model. This release provides practitioners with a powerful and efficient tool for generative AI applications, enhancing their ability to build and deploy large language models.