Anthropic, Google score win by nabbing OpenAI-backed Harvey as a user

Popular Legal AI Equipment Harvey Harvey will now use the top Foundation models from anthropologists and Google, OpenAI, Harvey is used strictly using it Declaration On a blog post on Tuesday.

This is noticeable because one of the most successful start-back portfolio companies in the Harvey Openi Startup Fund. OpenAI Startup Fund is a OpenAI Product developed companies at the top of AI Technologies are basically Openai’s own. Although Harvey says that it is not leaving the opening, only more models and clouds are adding, it is still a huge coup for the big competitors of the Open.

Harvey is one of the first four startups supported by OpenAI Startup Fund, it says At December 2022The Back OpenAI’s CEO Sam Altman Was still running funds. (That first team includes look, memes and spic in others)

Harvey, which has grown like crazy since then, is now the beginning of a 3 billion dollars-worth of value, it said in February, when it was Declaration A 300 million dollar series D, other big names such as Kotu, Clainer Parkins and OpenAI Fund Piling in the leadership.

The funny thing is, Google’s Venture Arm, GV, Leader Harvey is $ 100 million series c In July 2024 (and OpenAI Fund also participated in that round). Harvey Google’s AI models did not immediately accept the Harvey Google’s AI models after putting Google’s corporate venture firm on its cap table. (GV also participated in Harvey’s series D.)

So, did Harvey have persuaded the Openi models now? The internal developed benchmark of startup, known as Biglao, showed that different types of foundation models are increasingly growing in various legal purposes and some are better at certain tasks than others.

Instead of spending his effort training models, Harvey is depicted, it is just high-perf’s, rational foundation models reasonable (such as anthropological through Google and Amazon’s cloud) and then can make them fine for the legal market.

The company has said that the Harvey AI agent will help using different models as well.

Harvey Blog Post writes, “In less than a year, seven models (including three non-OI models) are now originally exceeded the benchmark Harvey system on the Bigla Bench,” wrote in the Harvey blog post.

Harvey’s benchmark also showed that different foundation models are better in certain legal work than others. For example, it says that Google’s Gemi 2.5 Pro is “Excels” in the Legal draft but there is a “struggle” with pre-judicial acts such as writing oral logic because the model does not fully understand “complex clear rules like hears.”

Openai and 3 do well pre-judicial works, according to Harvey’s test, ethnographic clad 3.7 Sonnet followed behind.

Harvey says in its blog post that it will join now A growing rank in sharing a public leaderboard Model Benchmark Performance. Its board will rank how big reasonable models are doing in legal activities. And the company will not only rank in a single number, but will also publish the research where the top lawyers provide short insights on model performances that are not captured by a single score benchmark. “

So, not only does the Open-backed Harvey competitors take the model, it is also increasing the pressure of its supporters (including Google) to prove them. Opeena’s score should not be too worried. When AI is benchmarking Growing complex and somewhat political growthIt is a world where the openness still shines.

Harvey CEO Winston Winston Weinberg told Techcunch in a statement, “We are incredibly lucky to be in the Harvey investor and Open as the main partner in our product.” “And, we are encouraged to add our options to customers as we are serving our customers worldwide.”