MS-SWIFT is a comprehensive LLM and multimodal model fine-tuning framework developed by Alibaba's ModelScope team, supporting over 600 text-only large language models and 300+ multimodal models. The framework provides a unified interface for the entire training lifecycle including continued pre-training (CPT), supervised fine-tuning (SFT), and preference learning via DPO, GRPO, and KTO algorithms. MS-SWIFT supports both full-parameter training and parameter-efficient methods such as LoRA, QLoRA, and DoRA, making it accessible for users with varying hardware budgets. For large-scale distributed training, it integrates Megatron-LM parallelism techniques including Tensor Parallelism (TP), Pipeline Parallelism (PP), Context Parallelism (CP), and Expert Parallelism (EP). The framework also includes a complete inference pipeline with vLLM, SGLang, and LMDeploy acceleration, plus model quantization support via GPTQ, AWQ, BNB, and FP8. Built on PyTorch with Flash-Attention 2/3 support and DeepSpeed optimization, MS-SWIFT runs on Python 3.10/3.11 and offers extensive documentation and training examples for researchers and practitioners.