# 面壁智能回应本地运行2B模型愿景

- 来源：OpenBMB (@OpenBMB)
- 发布时间：2026-09-08 22:23
- AIHOT 分数：34
- AIHOT 链接：https://aihot.news/items/cmtsrxgn1037troka0b5ra6t2
- 原文链接：https://x.com/OpenBMB/status/2097330090111881481

## AI 摘要

感谢测试并分享这些数据！🙌

“在本地运行一群这样的模型”——这正是我们的愿景。2B 参数，Q4 量化后 1.56GB，4090 上 200+ tok/s，同时性能仍能跟上 4B 模型。这就是本地 AI 智能体应有的样子。

期待看到你用它们构建的东西 🔥

## 正文

Thanks for testing it out and sharing these numbers! 🙌

"Run a swarm of these locally" — that's exactly the vision we had. 2B params, 1.56GB Q4, 200+ tok/s on a 4090, and still keeping up with 4B models. That's what local AI agents should look like.

Can't wait to see what you build with them 🔥 https://x.com/outsource_/status/2097005719983689998/video/1