# Google Gemma says vLLM can make DiffusionGemma act like Jev

Google Gemma says seeding a canvas with a response template in vLLM lets DiffusionGemma return confidence scores and probability distributions for yes/no, multiple-choice, and scored questions in a single denoising step.

Language: en
Time zone: America/Los_Angeles
Current time (UTC, cached as of): 2026-11-17T15:02:00.000Z
Query date (America/Los_Angeles): 2026-11-17

HTML: https://didcodexreset.com/news/f61f249f42444869893510fa.html

Content language: en
Localization state: sameLanguage

Source: Google Gemma · Published 9/29/2026, 10:47:56

Google Gemma says developers can turn DiffusionGemma into a Jev-like model for structured tasks.

By seeding a canvas with a response template, vLLM can extract confidence scores and probability distributions for yes/no, multiple-choice, and scored questions in a single denoising step.

Tags: DiffusionGemma, vLLM, Gemma, Jev

[View original post](https://x.com/googlegemma/status/2104990261181075498)
