About

Lexassay checks how well language models handle basic legal tasks, and whether the statutes and cases they cite are real and fit the claim.

What it is

A small, documented spot check. A fixed set of tasks with gold answers and primary sources is put to several models under the same conditions, without tools or web access. Every answer is published together with its grade and a check of each citation, so that readers can verify the results themselves. The method page explains the rules and the limits.

Who runs it

Lexassay is run by Thorsten Meyer in Iffeldorf, Germany, as part of Thorsten Meyer AI. It is independent: no model vendor sponsors, reviews or approves the results.

What it is not

Lexassay is not a law firm, gives no legal advice and is not a benchmark in the strict sense.

Get in touch

Questions, corrections and replies from model vendors: contact@lexassay.com.

Last updated: 2026-11-03