vLLM Adds Support for Gemma 4 Embedding and Diffusion Models
October 25, 2026
vLLM has integrated Gemma 4 support for EmbeddingGemma2 and DiffusionGemma via the _get_max_soft_tokens method. This update allows for improved soft token handling within the inference runtime for these specific model architectures.
HOW THIS AFFECTS YOU
●
builderYou can now deploy Gemma 4 embedding and diffusion models using the vLLM inference engine.
●
researcherThis improves the efficiency of handling soft tokens in Gemma-based architectures during inference.