Harvey builds its flagship legal model on Chinese open weights, not its investor's API
Tenet is post-trained on Moonshot's Kimi K3, converting per-call payments to OpenAI, Anthropic and Google into a fixed training bill. The counterparty risk moved rather than disappeared.
The Product Desk
What happened
- Harvey has launched Tenet, its first proprietary legal model, post-trained on Kimi K3, the open-weight model from Chinese startup Moonshot.
- Tenet is built to displace OpenAI's models inside Harvey's product, and OpenAI is one of Harvey's investors, alongside Sequoia and Andreessen Horowitz.
- The training material was manufactured: lawyers on staff and via Mercor and Snorkel wrote mock disputes and case files, then graded model reasoning on them.
- Harvey says the post-training was done with Fireworks AI.
Why it matters
- What it costs cost Harvey stops paying per call to companies whose products compete with its own and starts paying up front for training and legal data work it controls.
- What it limits constraint Independence from OpenAI's meter runs through Moonshot's licence: Harvey's revenue sits more than 17 times above the threshold at which that agreement has to be negotiated.
- Who's exposed exposure European law firms handling privileged matters take on a supplier whose base-model documentation duties sit with a lab in Beijing rather than with Harvey.
- What it sets up precedent Once a post-trained open base is deemed good enough for legal work, rivals such as Legora will be asked whose weights they are running as a sales question, not a technical one.
Next story · Invest · Sep 9
Harvey's $550M round prices legal AI at 39 times the ARR it just crossedConfirmed2 publishers
The base weights are the commodity in this arrangement. Anyone can download Kimi K3, released in July by Moonshot [2]. What cannot be downloaded is the grading: Harvey paid lawyers, some on staff and some through Mercor and Snorkel, to invent mock disputes and case files and then score how models reasoned through them [5]. That corpus exists because someone bought it, and it moves to whatever base looks best next year.
Cofounder Gabe Pereyra's description of the product, routing each task to whichever model suits it, is the doctrine rather than a hedge [9]. A system that already treats OpenAI, Anthropic and Google as interchangeable inputs [6] has defined its suppliers as swappable, and adding an owned option to that router is a procurement decision more than a research one.
The arithmetic makes the pressure legible. Harvey is valued at $15.5bn [8] on more than $350mn annualised [13], which puts the price at under roughly 44 times revenue [15]. A multiple in that range is a bet on margin expansion, and margin is hard to expand when the largest variable input is metered by firms selling into the same buying committee [4].
What the move does not remove is a counterparty. Kimi K3's licence permits derivative models but sends model-as-a-service operators above $20mn of revenue in any 12-month period to negotiate a separate agreement with Moonshot [12]. Whether a legal research product counts as model-as-a-service is the load-bearing definition here, and the source material does not settle it. If it does count, the negotiation is not a marginal one.
The regulatory result is stranger than the commercial one. Because a post-train sits well below the roughly one-third-of-original-training-compute marker in the Commission's guidelines, Harvey does not become the provider of the modified model, and most obligations stay upstream [11][17]. So Harvey gets the cost profile of ownership without the paperwork of authorship, and a European firm deploying Tenet inherits a documentation chain that begins in Beijing [17]. That is a manageable answer in a procurement questionnaire and a less comfortable one when the underlying work is privileged.
Legora, chasing a $10bn valuation and selling to many of the same firms [10], now has to answer the same question from the other side. In a market where Chinese weights are already free [18], the cheapest engine and the most awkward provenance are currently the same object.
What to watch
- Whether Harvey and Moonshot disclose a licence agreement, and whether Harvey's product is treated as model-as-a-service under it.
- Whether Harvey publishes evaluation results for Tenet against the OpenAI, Anthropic and Google models its router still calls.
- Whether Legora names its base model or commits publicly to frontier APIs as a differentiator with European buyers.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence55
- Adoption15
- Hype gap+30
- Incentives60
- Confidence55
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Harvey has launched Tenet, its first proprietary model for legal work, post-trained on Kimi K3, the open-weight model from Chinese startup Moonshot.
- [2]
Kimi K3 is an open-weight model released in July by the Chinese startup Moonshot.
- [3]
Harvey says the post-training work on Tenet was done with Fireworks AI.
- [4]
Every model call Harvey makes on a customer's behalf is an invoice from a rival, and owning the engine turns a variable cost into a fixed one.
- [5]
Harvey hired lawyers, on staff and through firms including Mercor and Snorkel, to invent mock disputes and case files and then grade how well models reasoned through them.
- [6]
For years Harvey routed customer work through models from OpenAI, Anthropic and Google.
- [7]
OpenAI is an investor in Harvey alongside Sequoia and Andreessen Horowitz.
- [8]
Harvey is valued at $15.5bn.
- [9]
Cofounder Gabe Pereyra, formerly of Google DeepMind, says Harvey already routes tasks to whichever model suits them, and Tenet adds an option shaped around legal work specifically.
- [10]
Legora, a Swedish competitor, is chasing a $10bn valuation and selling to many of the same firms as Harvey.
- [11]
Under the AI Act, a company that modifies someone else's general purpose model becomes the provider of the modified version only when the change is significant, with the Commission's guidelines using roughly a third of the original training compute as the indicative marker.
- [12]
The Kimi K3 licence permits derivative models but requires a separate agreement with Moonshot for model-as-a-service operators above $20mn of revenue in any 12 months.
- [13]
Harvey runs at more than $350mn annualised revenue.
- [14]
Harvey's annualised revenue is more than 17 times the $20mn licence threshold that triggers a separate agreement with Moonshot.
- [15]
Harvey's valuation implies a price of under roughly 44 times annualised revenue.
- [16]
Tenet is designed to displace OpenAI's models inside Harvey's own product.
- [17]
A post-train almost certainly falls well below the one-third compute marker, so most obligations stay upstream, and a European law firm deploying Tenet relies on a documentation chain that begins in Beijing.
- [18]
Harvey operates in a market already shaped by free Chinese models.
Sources
2 independent publishers whose own reporting we read for this story.
- Update on Harvey's Post-Training Effort
harvey.ai
1 article · August 25, 2026
- thenextweb.comHarvey’s first in-house model for legal work is here
1 article · August 23, 2026