SRI Lab, ETH Zurich
SRI Lab is the Secure, Reliable, and Intelligent Systems research group in ETH Zurich's Department of Computer Science. It researches the intersection of artificial intelligence and programming languages, with emphasis on secure, reliable, trustworthy, controllable and interpretable AI systems, and develops research systems and open-source tools.
At a Glance
- Academic and industrial AI/ML researchers
- Software developers building LLM applications
- Organizations needing reliable, secure and controllable AI systems
- AI model providers and deployers evaluating safety, security and EU AI Act compliance
- +2 more
AI Tools by SRI Lab, ETH Zurich
(1)
LMQL
Python Superset LLM Programming Language
Discussions
No discussions yet
Be the first to start a discussion about SRI Lab, ETH Zurich
Latest News
SRI team placed second in Track 3: SeePhys Pro at the 3rd AI for Math Workshop Challenge at ICML 2026.
SRI announced 8 papers in the main track of ICLR 2026, with two oral presentations.
SRI published research reporting that coding agents try to fix correct code over 50% of the time.
SRI work on the effect of context files on coding agents trended on X and Hacker News.
Products & Services
An open-source, Python-based programming language and platform for LLMs. It combines natural-language prompts with types, templates, constraints, control flow and an optimizing runtime to make LLM interaction more reliable, controllable, efficient and transparent. Website: https://lmql.ai/; repository: https://github.com/eth-sri/lmql
A high-level quantum programming language designed around intuitive semantics, reduced code, error prevention, safe automatic uncomputation and physicality. Website: https://silq.ethz.ch/
An open-source, compliance-centered evaluation framework for generative-AI models, created by ETH Zurich, INSAIT and LatticeFlow. It translates EU AI Act requirements into technical requirements and evaluates models using a 27-benchmark suite. Website: https://compl-ai.org/
Market Position
SRI is an academic research lab, not a conventional product company. It differentiates through the combination of programming-languages expertise, formal methods/program analysis, security and reliability, and applied generative-AI research, with open-source systems and spin-offs. Its research and talent landscape overlaps with university AI/programming-languages/security groups such as Berkeley BAIR, Stanford HAI, Mila and CMU, and industrial research organizations such as Google DeepMind, Microsoft Research and IBM Research.
Leadership
Executive Team
Martin Vechev
Full Professor of Computer Science, ETH Zurich; Head of SRI Lab
Leads SRI Lab. Previously Research Staff Member at IBM T.J. Watson Research Center in New York (2007–2011); PhD, University of Cambridge; BSc, Simon Fraser University. He co-founded DeepCode, ChainSecurity and LatticeFlow. Source: https://www.sri.inf.ethz.ch/people/martin
Founding Story
The lab grew from Martin Vechev's research at ETH on software reliability, programming languages, program analysis and AI for code, and expanded into reliable and secure AI/LLM systems. The group's stated work spans both theoretical and system aspects of AI and programming languages; ETH describes LMQL as an open-source effort created to make interaction with large language models more systematic, controlled and transparent.
Business Model
Revenue Model
SRI Lab is an ETH Zurich university research group rather than a separately marketed commercial company. Its public software projects, including LMQL, are open source; the lab's disclosed activity is research, teaching, grants and technology transfer/spin-offs rather than a public subscription or API business model.
Pricing Tiers
The LMQL website and GitHub repository provide installation, source code and a browser Playground; no paid tiers are listed. Repository license: Apache License 2.0.
Target Markets
- Academic and industrial AI/ML researchers
- Software developers building LLM applications
- Organizations needing reliable, secure and controllable AI systems
- AI model providers and deployers evaluating safety, security and EU AI Act compliance
- Quantum-computing researchers and developers
- Research and technology-transfer partners in cybersecurity, software engineering and trustworthy AI
- Controlled and constraint-guided LLM prompting
- Schema-safe structured generation and JSON-like output
- Tool-augmented, multi-step and interactive LLM applications
- Research and prototyping with local or hosted language models
- AI-for-code security hardening, code analysis and coding-agent evaluation
- LLM watermarking, privacy/security testing and reliable model evaluation
- No commercial customer list is published for SRI Lab. Publicly documented technology-transfer users/beneficiaries include SRI spin-offs DeepCode/Snyk, LatticeFlow, ChainSecurity/PwC, Invariant Labs/Snyk, LogicStar and NetFabric.