Maybe AI agents can be lawyers after all
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
Anthropic's Opus 4.6 model achieved a significant breakthrough on Mercor's professional task benchmark, jumping from 18.4% to nearly 30% accuracy on complex tasks like legal work and corporate analysis. The model scored 45% when given multiple attempts, representing a major leap in AI agent capabilities within just a few
Sort: