Ai Harvey’s popular legal tool will now use top foundation models by Anthropic and Google, moving beyond strictly using Openai, Harvey announced In a blog post on Tuesday.
This is remarkable because Harvey is one of the most successful portfolios of Openai Startup Fund. The OpenAi Starting Fund is an Openai -related fund for support companies that develop products above AI Technologies, notably Openai’s. While Harvey says he does not leave Openai, simply adding more models and clouds, this is still a huge coup for Openai’s great competitors.
Harvey is one of the first four newly established businesses to be supported by the OpenAi Basin Fund, he said In December 2022. This was back when Openai CEO Sam Altman was still running the fund. (Others in this first group include description, MEM and speak.)
Harvey, who has grown like crazy since then, is now a startup of $ 3 billion, he said in February when the announced A series of $ 300 million led by Sequoia, with other big names such as Coatue, Kleiner Perkins and the OpenAi fund that are accumulating.
Interestingly, Google’s Arm Venture Arm Arm Arm, GV, has led Harvey’s $ 100 million in July 2024 (and the Openai fund also participated in this round). But Harvey did not immediately adopt Google’s AI models, after putting Google’s business company on his table. (GV also participated in Harvey’s D row.)
So what did Harvey persuade to go beyond Openai models now? The internal developed reference point of the start, called Biglaw, showed that a wide variety of foundation models is increasing more and more experienced in a number of legal duties and some are better in specific tasks than others.
Instead of spending his efforts training, Harvey understood, he could simply embrace high performance, logic of other suppliers (eg Google and Anthropic through Amazon Cloud) and then refine them for the legal market.
Using a variety of models will also help as Harvey creates AI agents, the company says.
“In less than a year, seven models (including three non-oai models) are now overcoming the Harvey’s initial comparative system in Biglaw Bench,” Harvey writes in the blog post.
Harvey’s reference index also showed that different foundation models are better in specific legal duties than others. For example, he says that Gemini 2.5 Pro “superiorly” in legal pension, but “struggles” with pre-transcutation tasks, such as writing oral arguments, because the model does not fully understand “complex evidence such as listening”.
Openai’s O3 is doing well in such pre-transcutation tasks, according to Harvey’s tests, with the Sonnet of Claude 3.7 Anthropic after close.
In his blog, Harvey says he will also participate in the growing ranks of those who share a public leadboard of model reference performance. His Board of Directors will classify how important models of reasoning are made into legal duties. And the company will not only boil the ranking in a single number, but will also publish a survey where “top lawyers provide subtle knowledge of the performance of the model that has not been recorded by a score of reference”.
Thus, not only Harvey supported by OpenAI adopts competitors, but also pressure on its supporters (including Google) to continue to prove. Not that Openai should be very worried about this score. While AI’s comparative assessment is increasing more and more complex and somewhat political, this is a world where Openai is still shining.
“We are incredibly lucky to have Openai as a Harvey investor and a key partner in our product,” Harvey Winston Weinberg CEO told TechCrunch in a statement. “And, we are activated to add to our customers’ choices as we continue to serve the needs of our customers worldwide.”
