This is a promising but single-source customer case study indicating a material workflow improvement, not independent evidence of general-purpose model performance.
Limits of the evidence: The performance figures and planted-error results are reported by Legora in an official OpenAI customer case study; no independent evaluation or broader benchmark evidence is provided.
What to watch
Taken from the first report; the whole story has not been re-assessed yet.
Legora reviewed 41 documents in minutes with GPT-6 Astra
OpenAI NewsSep 3, 2026First report
No revisions yet.
No maturity ladder applies to this desk.
Sens.ai aggregates and assesses published reporting. The assessment above is machine generated; the original sources are authoritative.