← All projects

LMQL

A programming language for robust, modular LLM interaction

Dev Toolsllmprompt-engineeringpythonlanguage-modelconstraintsopen-sourceai-programming
LMQL screenshot

About

LMQL (Language Model Query Language) is a programming language designed for interacting with large language models. It allows developers to write prompts using Python control flow, typed variables, and hard constraints enforced at runtime. The project is created by the SRI Lab at ETH Zurich and supports multiple LLM backends.

Problem

Writing reliable, modular, and portable prompts for LLMs is difficult without type safety, constraints, or structured control flow.

For

software developers and researchers working with large language models

How it works

Developers write query functions in Python-like syntax with special prompt strings, typed variables, and constraint expressions that the LMQL runtime enforces during LLM generation.

Business model

open-source

Status

launched

Company

SRI Lab @ ETH Zurich

Similar projects