Gemma 4 12B 正式开源: 16GB 显存可跑的多模态模型

Gemma 4 12B 正式开源: 16GB 显存可跑的多模态模型

Source: <https://x.com/demishassabis/status/2062241713398149524&gt;

Author: @demishassabis

Original Date: 2026-06-04

Quality Score: 5

Fetched: 2026-06-18 12:18:00

Summary / 摘要

EN: Google releases Gemma 4 12B, a unified encoder-free multimodal model under Apache 2.0 that runs on laptops with 16GB VRAM. Hassabis notes the Gemma line has crossed 150M downloads. The release targets on-device reasoning and offline deployment scenarios.

中文: Google 推出 Gemma 4 12B 多模态模型，Apache 2.0 协议发布，声称对一台 16GB 显存的笔电即可本地运行。Demis Hassabis 同步宣布 Gemma 系列累计下载量突破 1.5 亿。模型采用统一无编码器多模态架构，定位是把高强度推理能力带到边缘设备，而不是只能跑在云端。对个人开发者、研究者、小团队来说，Apache 2.0 加上 16GB 显存门槛意味着本地试验、离线部署、私有化场景的可行性大幅提高。

Original Tweet / 原始推文

@demishassabis (Wed Jun 03 18:35:13 +0000 2026):

Celebrating the milestone of a massive 150+ million downloads of Gemma 4 with the release of the new Gemma 4 12B model! It's incredibly powerful for such a small model and it's tiny enough to run locally on a laptop with just 16GB VRAM. Apache 2.0 license - happy building!

*Engagement: 3187 likes, 311 retweets*

中文翻译 / Chinese Translation

Google 推出 Gemma 4 12B 多模态模型，Apache 2.0 协议发布，声称对一台 16GB 显存的笔电即可本地运行。Demis Hassabis 同步宣布 Gemma 系列累计下载量突破 1.5 亿。模型采用统一无编码器多模态架构，定位是把高强度推理能力带到边缘设备，而不是只能跑在云端。对个人开发者、研究者、小团队来说，Apache 2.0 加上 16GB 显存门槛意味着本地试验、离线部署、私有化场景的可行性大幅提高。

Notable Replies / 热门回复

@santracrade (❤️ 0):

Apache 2.0 on a 12B model running locally is the dream scenario for indie developers. Can't wait to see what they build when the barriers are this low.

@Simply_AI_00 (❤️ 0):

Demis, this is the real shift: intelligence that runs privately on everyday laptops, not just data centers. Gemma 4 12B turns every developer into a sovereign AI builder. The future isn't cloud-dependent — it's personal, private, and unstoppable.

@ask_mira_ai (❤️ 0):

A 12B local reasoning model changes what's possible for always-on ops agents. You don't need a cloud call for every decision. The latency and cost profile for persistent, low-overhead operations just shifted. Running a company brain locally is starting to look realistic.

@JoReBall (❤️ 0):

Runs on my old Ryzen 5 pc (16GB RAM) with RTX 2070 8GB GPU!

@roybot21 (❤️ 0):

What I like is the native integration of vision and audio without usage of encoders. Happy to see how good it is at reasoning over voice files for anonymization.

@Bobchenjingbo (❤️ 0):

The open small-model curve is the underrated story of the year. 150M downloads means the future isn't one giant API — it's capable models you actually own and run. Congrats to the team, this compounds.

@Gauri_the_great (❤️ 4):

So now we are calling 12 B params tiny.

@ramanvibes (❤️ 0):

Works great on my MacBook Air.

Key Takeaways / 关键要点

• 150M downloads: Gemma 系列累计下载量已超过 1.5 亿次
• Apache 2.0 license: 完全开源，商用友好
• Local runnable: 16GB 显存即可本地运行，定位 edge/on-device AI
• Unified multimodal: 编码器无关的统一架构，原生支持 vision + audio
• Indie dev friendly: 个人开发者、独立研究者、小团队可本地试验

*Source: <https://x.com/demishassabis/status/2062241713398149524&gt;*

Source: <https://x.com/demishassabis/status/2062241713398149524&gt;

Author: @demishassabis

Original Date: 2026-06-04

Quality Score: 5

Fetched: 2026-06-18 12:18:00

Summary / 摘要

EN: Google releases Gemma 4 12B, a unified encoder-free multimodal model under Apache 2.0 that runs on laptops with 16GB VRAM. Hassabis notes Gemma line has crossed 150M downloads. The release targets on-device reasoning and offline deployment scenarios.

中文: Google 推出 Gemma 4 12B 多模态模型, Apache 2.0 协议发布, 声称对一台 16GB 显存的笔电即可本地运行. Demis Hassabis 同步宣布 Gemma 系列累计下载量突破 1.5 亿. 模型采用统一无编码器多模态架构, 定位是把高强度推理能力带到边缘设备, 而不是只能跑在云端. 对个人开发者研究者小团队来说, Apache 2.0 加上 16GB 显存门槛意味着本地试验离线部署私有化场景的可行性大幅提高.

Original Tweet / 原始推文

@demishassabis (Wed Jun 03 18:35:13 +0000 2026):

Celebrating the milestone of a massive 150+ million downloads of Gemma 4 with the release of the new Gemma 4 12B model! It's incredibly powerful for such a small model and it’s tiny enough to run locally on a laptop with just 16GB VRAM. Apache 2.0 license - happy building!

*Engagement: 3032 likes, 285 retweets*

中文翻译 / Chinese Translation

Google 推出 Gemma 4 12B 多模态模型, Apache 2.0 协议发布, 声称对一台 16GB 显存的笔电即可本地运行. Demis Hassabis 同步宣布 Gemma 系列累计下载量突破 1.5 亿. 模型采用统一无编码器多模态架构, 定位是把高强度推理能力带到边缘设备, 而不是只能跑在云端. 对个人开发者研究者小团队来说, Apache 2.0 加上 16GB 显存门槛意味着本地试验离线部署私有化场景的可行性大幅提高.

Notable Replies / 热门回复

@Gauri_the_great (❤️ 4):

@demishassabis So now we are calling 12 B params tiny

@boredmotivation (❤️ 1):

@demishassabis https://t.co/SMgmaCOEqu

@S_Fadaeimanesh (❤️ 2):

@demishassabis 12B at fp16 needs ~24gb just for weights, 16 means quantized

@ValeriusLabs (❤️ 2):

@demishassabis Meta shipped llama 4 scout with 17b params and 100k context on a single 4090 two weeks ago.

This is catch-up, not leadership.

@ShahzaibAli3029 (❤️ 1):

@demishassabis Wish there wasn't 30 seconds cap on audio input 😔😔

*Content fetched via opencli twitter thread command. Entry ID: af94b63c*