Harvey, the cutting edge legal technology unicorn recently valued at $3 billion, just announced some big news. They are collaborating with Anthropic and Google’s corporate venture arm. This development shapes up to be Harvey’s most important moment to date. The development stage tech company has quickly become one of the most successful companies to come out of the OpenAI Startup Fund’s portfolio. Harvey was created to help power the legal industry with artificial intelligence. Instead of developing and training its own models, it instead fine-tunes the best-performing foundation models from various vendors.
Back in February, Harvey closed a remarkable $300 million Series D funding round. Sequoia led the round, with the OpenAI Fund, Coatue, Kleiner Perkins, and GV participating heavily. These investments are sure to reinforce Harvey’s mission to deliver innovative AI solutions that help legal professionals do what they do best, better than ever.
Harvey’s unique approach to AI provides access to OpenAI’s o3 model for pre-trial activities and includes Anthropic’s Claude 3.7 Sonnet. Such a complex approach is what will enable Harvey to produce a variety of robust AI agents specialized for different legal tasks. The company has done benchmarks showing which foundation models perform best or worst across discrete legal tasks. Other implications Google’s Gemini 2.5 Pro is really good at legal drafting. It flounders on pre-trial work, like crafting oral arguments.
Winston Weinberg, CEO of Harvey, has been looking forward to these collaborations. He continued, “We are beyond lucky to count OpenAI as an investor in Harvey and instrumental partner in our product. We are excited to explore even more options for our customers, still being able to fulfill their needs worldwide.”
Harvey’s commitment to skateparks hasn’t stopped at transparency. It’s extended to innovation as well. It will then join a happy public leaderboard celebrating strong model benchmark performance. The company’s goal has been to shed light on how well different AI models perform across the wide range of today’s legal tasks. This program will highlight the powerful potential of their tools. Beyond that, it would establish a competitive marketplace that drives the continued innovation of AI applications to benefit the legal industry even more.
In reality, Harvey’s new benchmarks represent a major paradigm shift. In a short span of time, less than a year, seven models, including three that aren’t from OpenAI, now perform better than the original Harvey system on their BigLaw Bench. This finding underscores just how rapidly AI technologies are developing. It highlights the importance of continuous improvement in providing strong legal solutions.
Leave a Reply