T
traeai
Sign in

产品

什么是 LiteRT

用于macOS桌面应用的本地推理引擎,支持Gemma 4 12B的高效端侧运行。

为什么现在值得关注?

最近变化

2026-06-03 · Gemma 4 12B通过统一架构移除独立多模态编码器,实现端到端多模态处理。

LiteRT 被反复提及时,通常意味着它正在影响产品路线、开发者工作流或 AI 产业判断。这个页面把分散材料合并成一个可持续更新的观察入口。

📰 LiteRT 最新动态

已收录 4 篇与「LiteRT」相关的 AI 资讯和分析。

Accelerating on-device AI: A look at Arm and Google AI Edge optimization

Accelerating On-Device AI: A Look at Arm and Google AI Edge Optimization

Google Developers Blog1644 字 (约 7 分钟)
85

The article introduces the collaboration between Arm and Google to optimize edge AI inference through the SME2 architecture and Google AI Edge toolchain.

入选理由:Arm SME2使CPU成为高性能AI加速器,推理速度提升5倍

FeaturedArticle#AI Edge#Arm#Machine Learning#Edge Computing中文
Blazing fast on-device GenAI with LiteRT-LM

Blazing fast on-device GenAI with LiteRT-LM

Google Developers Blog1574 字 (约 7 分钟)
75

Google AI Edge introduces LiteRT-LM, an optimized inference engine for deploying Gemma 4 models on edge devices, supporting Android, iOS, and web platforms with GPU inference reaching 76 tokens/sec and Multi-Token Prediction delivering up to 2.2x speedup.

入选理由:LiteRT-LM 在 Android GPU (OpenCL) 上实现 52 tokens/sec 解码速度,iOS (Metal) 达 56 tokens/sec,WebGPU 在 MacBook Pro 上可达 76 tokens/sec

FeaturedArticle#Google AI Edge#LiteRT-LM#Gemma 4#Edge AI#On-device Inference英文
Gemma 4 12B is here! 

It comes with a new, unified architecture that removes separate multimodal en...

Gemma 4 12B is here!

Patrick Loeber(@patloeber)172 字 (约 1 分钟)
72

Gemma 4 12B adopts a unified architecture removing separate multimodal encoders, enabling local vision/audio understanding and advanced agentic reasoning, with a new LiteRT-powered macOS desktop app.

入选理由:Gemma 4 12B通过统一架构移除独立多模态编码器,实现端到端多模态处理。

FeaturedTweet#Gemma 4#Multimodal LLM#LiteRT#Agentic AI英文
Google Developers Blog 图标

Google Tensor SDK Beta with LiteRT

Google Developers Blog959 字 (约 4 分钟)
65

Google releases Tensor SDK Beta for on-device ML on Pixel 10 devices, featuring unified workflow with LiteRT and a Model Garden of 100+ models supporting PyTorch/TFLite compilation and TPU inference deployment.

入选理由:Tensor SDK Beta现已支持Pixel 10系列设备,可调用Tensor SoC中专用TPU进行推理加速

FeaturedArticle#Google Tensor#LiteRT#Edge AI#On-device ML#Pixel英文

与「LiteRT」经常一起出现的 AI 术语。

💡 想追踪「LiteRT」的长期趋势?去 实体雷达 · LiteRT 查看详细分析和跨材料问答。

AI may generate inaccurate information. Please verify important content.