26B MOE 最近有什么新动态？

traeai 已收录 2 篇与 26B MOE 相关的内容。最新一篇是「Gemma-4 12B + Hermes,Google AI Edge: EASY, GOOD & LOCAL!」，由 AICodeKing 发布。

模型

26B MOE

别名：26B Mixture of Experts

Gemma 系列中的一种混合专家模型，作为性能基准对比对象。

Gemma-4 12B + Hermes, Google AI Edge: EASY, GOOD & LOCAL!

AICodeKingYesterday3109 字 (约 13 分钟)

Gemma-4 12B is an encoder-free, unified multimodal model that runs directly on laptops with 16GB VRAM. It matches the performance of the 26B MOE with less than half the memory footprint, ships with Hermes and agent tools, macOS Edge Gallery, and RTLM, and is released under Apache 2.0.

入选理由：Gemma-4 12B 无需分别的视觉/音频编码器，图像与音频直接映射到 LLM，减少延迟与内存开销。

FeaturedVideo#Gemma#412B#Multimodal#Local Deployment#Hermes英文

Introducing Gemma 4 12B: a unified, encoder-free multimodal model

The Keyword (blog.google)Yesterday693 字 (约 3 分钟)

Gemma 4 12B is a unified, encoder-free multimodal model bringing high-performance multimodal intelligence to your laptop. It matches the performance of our 26B MoE at less than half the memory footprint, supports native audio inputs, and runs locally on 16GB VRAM hardware with low-latency multi-step reasoning.

入选理由：Gemma 4 12B 性能接近 26B MoE，内存仅其一半，适合在 16GB VRAM 现代本机运行。

FeaturedArticle#Gemma 4#12B#multimodal#unified architecture#encoder-free英文

跨材料问答 · 26B MOE

回答基于：26B MOE 相关 2 条材料

26B MOE

相关材料

Gemma-4 12B + Hermes, Google AI Edge: EASY, GOOD & LOCAL!

Introducing Gemma 4 12B: a unified, encoder-free multimodal model

跨材料问答 · 26B MOE