Phi-3 Mini
A small language model developed by Microsoft, designed for deployment on edge devices. It features a low parameter count and high inference efficiency, enabling local execution on resource-constrained mobile devices.
Timeline (last 90 days)
Phi-3 Mini, Gemma 2B, and Llama 3.2 1B are representative small language models (SLMs), with parameter counts typically ranging from 1B to 7B.
Mainstream on-device large models (such as Gemini Nano and Phi-3 Mini) typically require at least 6–12 GB of RAM to run smoothly.
Llama 3.2 3B、Phi-3 Mini 3.8B、Gemma 2B 等小模型在简单问答、文本摘要和工具调用任务上具备实用能力，但在复杂推理、长文本理解和多轮对话一致性上仍与 GPT-4、Claude 等云端旗舰模型存在明显差距
典型的端侧LLM如Google Gemini Nano、微软Phi-3 Mini参数量通常控制在1B-4B范围内
All Facts (4)
Phi-3 Mini, Gemma 2B, and Llama 3.2 1B are representative small language models (SLMs), with parameter counts typically ranging from 1B to 7B.
50%UnverifiedMainstream on-device large models (such as Gemini Nano and Phi-3 Mini) typically require at least 6–12 GB of RAM to run smoothly.
50%UnverifiedLlama 3.2 3B、Phi-3 Mini 3.8B、Gemma 2B 等小模型在简单问答、文本摘要和工具调用任务上具备实用能力，但在复杂推理、长文本理解和多轮对话一致性上仍与 GPT-4、Claude 等云端旗舰模型存在明显差距
50%Unverified典型的端侧LLM如Google Gemini Nano、微软Phi-3 Mini参数量通常控制在1B-4B范围内
50%