DSPy

DSPy is a powerful open-source framework for programming—rather than prompting—language models. Created at Stanford NLP, DSPy provides a systematic way to compile declarative language model calls into self-improving, optimized pipelines. With 35k+ GitHub stars, it enables developers to build sophisticated LLM applications by automatically optimizing prompts, chain-of-thought reasoning, and model interactions. DSPy abstracts away the fragility of hand-crafted prompts, replacing them with a compiler that learns from data to produce the best program for each task. It supports all major LLM providers, integrates with retrieval pipelines, and offers modular building blocks for classification, generation, retrieval-augmented generation (RAG), multi-hop QA, and agent-style tool use. DSPy is an essential toolkit for any developer building production LLM applications.