Harvey Moves Flagship Off Frontier Labs to Moonshot's Kimi K3

Harvey's gross margins collapsed from about 50% to minus 50% by June as token consumption jumped twentyfold under OpenAI and Anthropic's usage-based enterprise pricing, Bloomberg reported. Margins turned positive again after the legal AI company launched its own in-house model in August, post-trained on Moonshot's open-weight Kimi K3.

Harvey Tenet, released last month, sits on the Chinese base rather than a frontier US model. In the South China Morning Post's telling, the San Francisco startup "previously focused on customising closed proprietary models from US leaders such as Anthropic, OpenAI and Google for legal applications." Its investors include the OpenAI Startup Fund, Sequoia Capital and Andreessen Horowitz. A company valued at $15.5bn has moved its flagship onto weights released in July by Chinese lab Moonshot, in work Harvey says it did with Fireworks AI.

Bloomberg names Abridge, Decagon and Ramp as pursuing similar open-weight pivots, backed by Sequoia Capital and General Catalyst. The story lands in the middle of a run of Anthropic pricing and OpenAI capital coverage we've been tracking day after day this month.

Neither Bloomberg nor SCMP publishes Harvey's post-Tenet margin number, or what share of its workload still routes to GPT and Claude for the hardest matters.