arXiv 2510.03847
Small Language Models for Agentic Systems: A Survey of Architectures, Capabilities, and Deployment Trade offs
By Raghav Sharma and Manan Mehta
Published 2025-10-04
Citation lineage
Review the prior work and downstream research connected to this paper.
Small language models (SLMs; 1-12B params, sometimes up to 20B) are sufficient and often superior for agentic workloads where the objective is schema- and API-constrained accuracy rather than open-ended generation. We synthesize recent evidence across open and proprietary SLMs (Phi-4-Mini, Qwen-2.5-7B, Gemma-2-9B, Llama-3.2-1B/3B, Ministral-3B/8B, Apple on-device 3B, DeepSeek-R1-Distill) and connect it to modern eva…