Outlines
Guaranteed structured generation with regex, JSON Schema and grammars.
Constrains token sampling so outputs always match a JSON schema, regex or context-free grammar; works with transformers, vLLM, llama.cpp and hosted APIs.
- Vendor
- .txt (dottxt)
- Category
- LLM & RAG frameworks
- Pricing
- Open source
- Open source
- Yes
- License
- Apache 2.0
- Platforms
- Python
- Launched
- 2023
- Website
- dottxt-ai.github.io/outlines
Features
- JSON Schema / Pydantic
- Regex & CFG constraints
- Many backends
Best for
- Extraction & classification
- Agents
More llm & rag frameworks
- LangChain — LangChain. The most widely used LLM application framework.
- AI SDK — Vercel. The TypeScript toolkit for AI apps.
- LlamaIndex — LlamaIndex. Data framework for RAG and document agents.
- DSPy — Stanford NLP. Program — don’t prompt — language models.
- Haystack — deepset. Production-ready RAG pipelines.
- Instructor — Jason Liu & community. Pydantic-typed structured outputs from any LLM.
- BAML — Boundary. A typed language for defining and testing LLM functions.