Harvey's first LLM for legal work is here
Harvey launched Tenet, its first proprietary large language model built specifically for legal work, designed to handle tasks that typically take lawyers hours or days.
Harvey has launched its first in-house, proprietary large language model, called Tenet, built specifically for legal work, according to Business Insider. The announcement marks the legal tech company's first purpose-built AI model, moving beyond fine-tuned versions of third-party foundation models.
Tenet is designed to take on legal tasks that attorneys would otherwise spend hours or days completing, per Business Insider. The model is trained specifically for the demands of legal work rather than general-purpose use, reflecting a broader push in legal tech to develop domain-specific AI rather than adapting general models to professional contexts.
Harvey, co-founded by Gabe Pereyra, has established itself as one of the more prominent AI companies focused on the legal sector. The company has previously relied on underlying models from outside providers to power its legal AI platform, making Tenet a notable internal milestone — the first model developed and owned entirely in-house.
The development of a proprietary model signals Harvey's ambition to differentiate its platform at the model layer, where legal-specific training data and reasoning capabilities can offer advantages that general-purpose LLMs are unlikely to close quickly. Law is a field where precision, citation, and jurisdiction-specific reasoning matter considerably, creating a meaningful surface area for domain-specialized models to outperform their general counterparts.
Details on Tenet's specific capabilities, benchmark performance, or rollout timeline to existing Harvey customers were not included in the available reporting from Business Insider. Further technical specifics, pricing changes, or customer availability details may follow as Harvey makes broader announcements around the launch.
No comments yet — start the thread.