Unverified50% confidenceFactExact time
The first 20 layers of the Gemini Nano model share the KV cache and support a context length of 16K tokens.
1
Sources
50%
Confidence
Long-term
Relevance
10/29/2026
First Seen
Sources
Android ML Kit On-Device AI: Performance Optimization and Quality Breakthroughs
youtubeAndroid Developers10/29/2026
Related Entities
Related Claims
UnverifiedGemini Nano是专为移动端设计的轻量版本，参数规模通常在18-32.5亿参数范围，在Pixel设备上通过Tensor芯片内置的TPU运行76% similarUnverifiedGemini offers explicit context caching, billed by storage duration and token volume.73% similarUnverifiedThere are rumors that the new Gemini 4 Pro model may support a context length of up to 2M tokens.72% similarUnverifiedGemini Nano是参数量在1.8B至3.25B之间的端侧小模型，支持断网状态下执行语音指令理解、导航意图识别等核心任务71% similarUnverifiedGemini Nano has undergone quantization and knowledge distillation processing, enabling inference on consumer-grade CPUs or mobile NPUs.70% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/1112919API
curl https://kongchang.com/api/v1/knowledge/claims/1112919MCP
get_claim(id=1112919)