Harvey pricing

About

Harvey is a generative-AI platform built for legal and professional-services work — drafting, contract analysis, legal research, and multi-step workflow automation across its Assistant, Vault, Knowledge, Workflows, and Word Add-In products. It targets the world’s most demanding legal teams, and its public customer roster (shown on the demo page) includes large law firms and enterprises such as Gleiss Lutz, Vinson & Elkins, Schönherr, Deutsche Telekom, KKR, and Bridgewater. Co-founder Winston Weinberg is a former litigator at O’Melveny & Myers, and that practitioner credibility helped Harvey reach more than 100,000 lawyers across 1,300+ organizations, including 45+ AmLaw 100 firms.

By the numbers, Harvey is one of the fastest-scaling vertical-AI companies on record: it crossed $100M ARR in August 2025 (roughly three years after founding), reported approximately $190M–$195M ARR by early 2026, and raised $200M at an $11B valuation in March 2026 co-led by GIC and Sequoia — up from an $8B valuation only three months earlier (per CNBC and Sacra). It competes most directly with Thomson Reuters CoCounsel and an emerging field of legal-AI challengers.

Harvey sells exclusively through an enterprise, sales-led motion. As captured on 2026-05-30, the company publishes no pricing page — harvey.ai/pricing returns HTTP 404 and there is no “Pricing” or “Plans” link anywhere in the site navigation. The only commercial surface is a “Book a demo” / contact-sales form. Every figure in the sections below that carries a dollar amount is a third-party estimate, clearly attributed and explicitly not confirmed by Harvey — which has publicly disputed such estimates.

Pricing summary : How Harvey’s pricing model works

Harvey’s pricing is fully gated. There is no public rate card, no self-serve checkout, and no free tier — the only path to a price is to book a demo through harvey.ai/contact-sales and receive a custom enterprise quote. No monthly price, per-seat rate, or usage unit is disclosed anywhere on the public site as of 2026-05-30. The model is best classified as per-seat subscription on annual commitments, sold through a sales-led motion.

• Sales motion: enterprise, sales-led only (demo → quote). No PLG or self-serve path exists.
• Published price: none. harvey.ai/pricing returns a 404; the public site is a demo-request funnel.
• Reported (not confirmed) billing dimensions: per-seat, annual term, with seat minimums. Third-party sources cite roughly 20–25 seat minimums and conflicting per-lawyer figures (some reports quote a monthly four-figure rate, others a yearly four-figure base) — estimates that disagree with each other and that Harvey has called “wildly off.” See the reported-figures table under Hidden costs for the attributed numbers.

What makes this different: Unlike many AI tools that lead with a transparent self-serve tier, Harvey runs a 100% quote-based enterprise motion with zero price transparency — and actively contests the third-party numbers that circulate in its place, a posture few peer vendors take publicly.

Pricing by product

Harvey does not break pricing out by product. The platform spans several products — Assistant, Vault, Knowledge, Workflows, and a Word Add-In — but all are sold together under a single enterprise agreement with no per-product price published.

Harvey platform (the only plan)

Sales motions across products: sales-led only — every product is bundled into one quoted enterprise agreement. There is no PLG / self-serve motion at Harvey. For the conflicting third-party per-seat estimates, see Hidden costs below — none are Harvey-confirmed.

Hidden costs : What Harvey users actually pay

Because Harvey discloses no rates, the “hidden costs” of a Harvey deal are best understood as the structural commitments a gated, seat-minimum, annual-contract model forces — not line-item overages. Every dollar figure in this section is a third-party reported estimate, not a Harvey quote, and the estimates conflict with one another.

Reported per-seat figures (third-party estimates — NOT confirmed by Harvey)

The only numbers in circulation come from third parties; Harvey publishes nothing and has publicly called such estimates “wildly off.” Treat this as reported market chatter, not vendor pricing:

The order-of-magnitude disagreement between the “per-year” and “per-month” camps is itself the headline finding: nobody outside Harvey’s sales process knows the real number, which is exactly the position gated pricing is designed to create. The June 2025 LexisNexis content partnership is expected by third parties to push all-in seat cost up (one editorial projected a roughly 30–40% uplift on 2026 contracts) — a projection Harvey disputes.

Modeled bills (scenarios from reported estimates — not quotes)

Archetype 1 — a 25-attorney mid-market firm adopting Harvey at the reported low end. Using the frequently-cited ”~$1,200 per lawyer per month” figure and a ~25-seat minimum:

The lesson: even at the reported low end, the 20–25 seat minimum is the real hidden cost — a firm cannot buy three seats to pilot, so the entry ticket is six figures annually before any usage discussion.

Archetype 2 — an AmLaw 100 deployment at the reported high end. Using the “$1,500–$2,000+ per seat/month” figures circulating for large firms across, say, 100 attorneys:

The lesson: the second hidden cost is content economics. Since June 2025 Harvey reportedly pays LexisNexis per retrieval, shifting some of its own cost base to usage — which third parties expect it to pass through as higher seat prices over time, even though the customer still sees a flat per-seat bill. See why usage-based input costs reshape pricing and the move away from flat per-user licenses.

Want to estimate your own Harvey bill? Use the Harvey pricing calculator to model scenarios across seat count and reported per-seat ranges — remembering that every figure is a third-party estimate, not a Harvey quote.

Pricing evolution : Harvey pricing history and changes

Harvey has never published a rate card, so there is no public price history to trace — the pricing surface has been gated since launch. What is observable is the structural and content-bundling activity that moves Harvey’s underlying cost base and, by extension, its negotiated seat prices.

Cadence

Tracked range: 2025–present. Harvey publishes no rate card, so per-quarter price values are recorded as unknown rather than estimated. Only structural/content events are dated.

Notable changes

• 2025-06-18 — Harvey announces a LexisNexis partnership bundling primary-law content; Artificial Lawyer estimated a meaningful all-in seat-cost increase, and Harvey publicly disputed the figures as “wildly off.”
• 2025-08 — Harvey crosses $100M ARR roughly three years after founding (per Sacra).
• 2025-12 — A16Z-led round at an $8B valuation ($160M raised), per TechCrunch.
• 2026-03-25 — $200M growth round at an $11B valuation co-led by GIC and Sequoia (CNBC). None of these milestones came with any public pricing disclosure.

The “wildly off” pricing dispute in detail

In June 2025, Artificial Lawyer modeled how the LexisNexis content bundle might reshape Harvey’s per-seat economics, projecting that all-in seat cost could climb by roughly one-third (≈$400–$600 per lawyer per year) and that a premium tier could approach ~$3,000 to stay below CoCounsel + Westlaw. Harvey responded on the record that the assumptions were “wildly off,” while still declining to share actual rates. This exchange is the single most concrete public signal about Harvey pricing — and it confirms the gating thesis: the only thing Harvey will say about its prices is that the public estimates are wrong.

What’s unique : Harvey’s distinctive pricing mechanics

1. Zero price transparency, by design. Harvey publishes nothing — not a tier list, not a “starting at” anchor, not a contact-sales table with a hidden range. harvey.ai/pricing is a 404 and the commercial surface is a bare demo form. This is more opaque than most sales-led enterprise vendors, which usually expose at least a published “Enterprise — contact us” tier alongside self-serve plans.

2. Per-seat economics layered over usage-based input costs. The customer sees a flat per-seat subscription, but since the June 2025 LexisNexis deal Harvey reportedly pays per content retrieval underneath. That hybrid — flat price to the buyer, metered cost to the vendor — is a classic margin-management pattern that pushes seat prices up over time without ever exposing usage to the customer.

3. Actively contesting third-party pricing estimates. Most gated vendors stay silent when bloggers reverse-engineer their prices. Harvey instead went on record calling Artificial Lawyer’s estimates “wildly off” — a deliberate move to suppress price anchoring in the market while still revealing nothing concrete.

4. Six-figure entry ticket via seat minimums. Reported 20–25 seat minimums on annual terms mean there is no small-pilot path: the smallest realistic deal is roughly $300K/year even at the reported low end. The minimum is the moat — it filters out everyone but firms that can commit a full practice group.

5. Premium positioning anchored to substitution cost. Reported high-end figures (~$1,500–$2,000+/seat/month) are positioned to sit just under the combined cost of incumbent stacks like CoCounsel + Westlaw, so the pitch is “cheaper than the legacy bundle, far more capable” — value framed against substitution rather than against a published list price.

Strengths & weaknesses

Billing UX : Harvey billing controls and transparency

• No self-serve billing surface — there is no public sign-up, checkout, or account/billing portal exposed; the only commercial control on the public site is the “Book a demo” form at harvey.ai/contact-sales.
• Demo-request funnel — the contact page is a single lead-capture form (it failed to load client-side in our capture, showing “Unable to load form. Please try again.”), with no pricing, plan selector, or seat calculator.
• Customer-proof, not pricing — the demo page surfaces enterprise logos (Gleiss Lutz, Vinson & Elkins, KKR, Bridgewater, Deutsche Telekom) in place of any rate card, signalling an enterprise-trust motion rather than transparent billing.
• In-product billing controls — any usage dashboards, spend caps, or invoicing terms live behind the enterprise agreement and are not observable publicly; left for the research stage.

Strategic wins : Why Harvey’s pricing decisions worked

1. Gating pricing to protect premium positioning

By publishing no rate card, Harvey avoids anchoring against cheaper legal-AI challengers and keeps every negotiation on a value-vs-substitution-cost footing rather than a feature-for-dollar comparison. For a category where the alternative is the expensive CoCounsel + Westlaw bundle, “cheaper than your legacy stack” is a stronger frame than any list price. See usage-based pricing strategy for why anchoring control matters.

2. Per-seat simplicity over metered complexity

Harvey could have exposed token or query metering to customers; instead it sells a flat per-seat subscription and absorbs the usage volatility itself. Lawyers hate unpredictable bills, and a flat seat cost is far easier for a firm’s finance team to approve than a metered AI bill. This is the opposite bet from the outcome- and usage-based trend — and for legal buyers it works.

3. Seat minimums as a qualification filter

Reported 20–25 seat minimums on annual terms turn pricing into a sales-qualification tool: the floor price is high enough that only serious, full-practice-group buyers engage, which keeps Harvey’s expensive sales-led motion focused on deals worth closing. It is a deliberate trade of TAM breadth for deal quality and ACV.

4. Owning the pricing narrative by rebutting estimates

Rather than letting third-party numbers calcify into “the price of Harvey,” the company publicly contests them. Calling estimates “wildly off” costs nothing, reveals nothing, and prevents competitors from undercutting a number Harvey never actually set. See choosing the right usage metric for how value framing beats published rates in enterprise sales.

Areas to improve : Gaps in Harvey’s pricing approach

1. Publish a directional anchor for mid-market firms

Total opacity filters out smaller firms that could become future AmLaw customers but won’t engage sales without a ballpark. A published “Enterprise starts at ~$X/seat, N-seat minimum” range — even a wide one — would qualify mid-market interest earlier without sacrificing negotiating room, and would blunt the conflicting third-party estimates that currently fill the vacuum. See bill shock and cost unpredictability for why buyers fear the unknown more than the high.

2. Offer a real pilot path below the seat minimum

The reported 20–25 seat floor means there is no low-commitment way to validate Harvey on a single matter or practice group. A capped, time-boxed pilot tier (even at a premium per-seat rate) would convert curiosity into adoption and de-risk the six-figure annual decision for cautious GCs.

3. Make content-cost pass-through legible

If LexisNexis per-retrieval economics are driving seat prices up, customers will eventually want to understand what they are paying for. A transparent (even if optional) usage view showing retrieval volume would pre-empt the “why did renewal go up 30%?” conversation. See the shift away from flat per-user licenses for how hidden usage costs erode trust at renewal.

Key takeaways

1. Opacity is a viable strategy in high-trust enterprise categories. When buyers are large firms negotiating per-deal and the alternative is an expensive incumbent bundle, a published price helps competitors more than customers. Harvey shows you can scale to $190M+ ARR with zero public pricing.
2. Flat per-seat pricing beats metering when the buyer fears unpredictability. Legal finance teams approve a flat seat cost far more readily than a metered AI bill. Absorbing usage volatility vendor-side is a feature, not a cost.
3. Seat minimums double as a sales-qualification filter. A high floor (~20–25 seats) keeps an expensive sales motion pointed only at deals worth closing, trading TAM breadth for ACV and deal quality.
4. Own your price narrative even when you publish nothing. Publicly rebutting third-party estimates (“wildly off”) costs nothing and prevents a number you never set from becoming “your price” in the market.
5. Premium positioning works best when anchored to substitution cost. Pricing just under CoCounsel + Westlaw lets Harvey frame value as “cheaper than your legacy stack, far more capable” — a stronger frame than any list price comparison.

UBP implications

1. Usage can live on the cost side without touching the customer’s bill. Harvey’s reported per-retrieval LexisNexis economics show a hybrid where the vendor is metered but the buyer pays a flat seat price — usage-based pricing as a margin tool, not a billing model.
2. Not every AI category should expose metering to buyers. For risk-averse, high-trust verticals like legal, flat per-seat predictability is a stronger value-metric choice than tokens or queries — a reminder that the right usage metric is buyer-dependent.
3. Vendor-side usage costs eventually surface as price pressure. When input costs are metered (per-retrieval content), seat prices drift upward at renewal even under a flat model — UBP practitioners should plan for that pass-through and the trust conversation it triggers.

Sources

• Harvey official website (accessed 2026-05-31)
• Harvey “Book a demo” / contact-sales page (accessed 2026-05-31)
• Harvey platform overview (accessed 2026-05-31)
• Harvey blog — $11B valuation growth round (accessed 2026-05-31)
• harvey.ai/pricing returns HTTP 404 — no public pricing page exists (verified 2026-05-30 via live capture).

Reported per-seat figures, funding/ARR milestones, and the LexisNexis pricing dispute are sourced from third-party press and analysis (CNBC, TechCrunch, Sacra, Artificial Lawyer, eesel AI) and are cited inline in the relevant sections. They are reported estimates, not Harvey-confirmed pricing.

Bottom line

Harvey is the breakout legal-AI platform — $190M+ ARR, an $11B valuation, and 45+ AmLaw 100 firms — sold entirely behind a gate. It publishes no rate card, runs a 100% sales-led per-seat motion with reported 20–25 seat minimums, and actively rebuts the third-party estimates ($1,000–$2,000+/lawyer figures) that circulate in place of a price. The pricing story isn’t a number; it’s the deliberate absence of one, paired with premium positioning under the incumbent CoCounsel + Westlaw bundle. Every dollar figure here is a reported estimate, not a Harvey quote.

Want to compare Harvey against other gated, sales-led AI vendors? Browse the pricing blueprint.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

No public pricing surface — fully gated

May 2026

Live capture confirms Harvey publishes no rate card: harvey.ai/pricing returns HTTP 404 and the only commercial surface is a book-a-demo form at harvey.ai/contact-sales showing enterprise logos in place of any price. No per-seat or usage rate appears anywhere on the public site.