Retired

This documentation is retired

The LLM evaluation platform documented on this site is retired and no longer offered. Its evaluation API, SDKs, evaluators, Lynx, GLIDER and Percival are all retired.

Patronus AI is a research lab developing training data and simulated environments for frontier AI. Its work spans reinforcement learning environments and digital world models for agents operating in software.

Public research

• FinanceBenchOpen-book financial question answering about publicly traded companies. 10,231 questions with answers and evidence strings.
  Dataset →
• GLIDERA small, explainable judge model for grading LLM interactions. Retired as a product, the paper stays public.
  Paper →
• LynxAn open-source model for detecting hallucinations in retrieval-augmented generation. Retired as a product, the paper stays public.
  Paper →
• ResearchRecent papers and earlier work from Patronus AI.
  Research →

Retired products

• EvaluatorsThe evaluators and evaluation API are retired.
  Retired →
• PercivalPercival is retired. Related research is TRAIL.
  Retired →
• SDKsThe Patronus SDK needs an API key, and new users cannot sign up.
  Retired →

Patronus AI

Founded

2023 by Anand Kannappan and Rebecca Qian

Headquarters

San Francisco

Website

www.patronus.ai (opens in a new tab)

• What is Patronus AIA behavioral research lab for AI.
  Company →

Clients with a contract have an offline copy of the former documentation. For questions, contact support@patronus.ai.