Google DeepMind releases Gemma 4 12B: a unified encoder-free multimodal model
Original title:Introducing Gemma 4 12B: a unified, encoder-free multimodal model
AI summary
Gemma 4 12B is Google DeepMind's latest medium-scale multimodal model, featuring an encoder-free unified architecture with native audio input support. Its benchmark performance approaches that of the 26B MoE model while using less than half the memory, running locally on consumer laptops with only 16GB of VRAM or unified memory. The model includes a built-in multi-token prediction (MTP) drafter to reduce latency, is released under the Apache 2.0 open-source license, and has surpassed 150 million downloads.
Key Takeaways
- Gemma 4 12B is Google DeepMind's latest medium-scale multimodal model,featuring an encoder-free…
- Its benchmark performance approaches that of the 26B MoE model while using less than half the…
- The model includes a built-in multi-token prediction (MTP) drafter to reduce latency,is released…
- 0 open-source license,and has surpassed 150 million downloads.
Quick Facts
- Published
- Updated
- Source
- Google DeepMind:Blog(RSS)
- Category
- Model Releases
- AI score
- 80
统一无编码器架构让 12B 模型在消费级笔记本上跑出接近 26B 的多模态 Agent 体验,开源 + Apache 2.0,本地部署门槛又压低了。
Related entities
Source & copyright
This article is based on curated titles and AI summaries from the AIHOT public API, with attribution and links back. Original copyright remains with the original publishers and authors; this site does not mirror full texts.
How to Cite
When citing this article, keep the title, date, and permalink, and credit Orange News as the source.
Orange News: Google DeepMind releases Gemma 4 12B: a unified encoder-free multimodal model. June 3, 2026. https://new.nat6.net/en/news/cmq6qqqbp0axhsl5i2ow5j5co/