What matters in AI.

Grok 4.7 is first in the new legal AI test with 9.4%

Artificial Analysis says a hallucination is material if it can mislead a reader, for example with an incorrect contract date.

Artificial Analysis and Harvey announced version 1.1 of the score method for the Legal Agent Benchmark. The new method adds a hallucination check. A task counts only if the work follows all the rubric criteria and has no material hallucination. Grok 4.7 from SpaceXAI is first with 9.4%. Artificial Analysis says that more than 60% of the results that are correct without the check contain a material hallucination.

Sources

@ArtificialAnlys: Today we are announcing Harvey LAB-AA v1.1 in collaboration with Harvey. This updates our scoring methodology for the Legal Agent Benchmark (LAB) to add a hallucination check and reqx.com

Posted 9 October 2026, 1:00 am IST

Learn the terms in this story

Guides for this story

Grok 4.7 is first in the new legal AI test with 9.4%

Sources

More in Benchmarks

Mathematicians find a mismatch in the OpenAI Navier-Stokes proofs

TestJack finds that 34.4% of trials shown as correct are incorrect

Reruns of the same tasks change 15 to 36 percent of the results