Comparisons
Harvey vs Legora: The Two Legal AI Platforms Compared
Harvey vs Legora in September 2026: $15.5bn vs $5.6bn valuations, ARR, clients, Workflow Agents vs Agentic OS, models, pricing minimums, geography and who should pick which.
On this page
Harvey is worth $15.5bn after a $550m round on 9 September 2026; Legora is worth $5.6bn and reportedly in talks at $10bn-plus. Harvey is the safer pick for a US-centred firm with 20 or more seats and a large in-house clientele; Legora is the stronger entry point for a European firm that wants Tabular Review, a 10-seat minimum and a Stockholm-based vendor. Both run the same frontier models.
The two companies in numbers
| Measure (as of Sep 2026) | Harvey | Legora |
|---|---|---|
| Founded / HQ | 2022, San Francisco; CEO Winston Weinberg | 2023, Stockholm; CEO Max Junestrand |
| Latest valuation | $15.5bn (9 Sep 2026; Lightspeed, Diffusion) | $5.6bn (29/30 Apr 2026; Atlassian, Nvidia's NVentures); reportedly seeking $10bn+ (Aug 2026, unverified) |
| Total raised | More than $1.5bn since 2022 | About $800m |
| ARR | Above $400m | $100m+ (Apr 2026); reportedly ~$150m (Q2 2026) |
| Customers | 3,000+; 80% of the top 100 law firms; 20% of the Fortune 500 | 1,480+ |
| Staff | Not disclosed | 675+ |
| ILTA 2026 share of firms | 43% overall; 84% of firms with 700+ lawyers | 16% |
| Acquisitions in 2026 | Hexus (Jan) | Walter AI (13 Mar), Qura (23 Apr), Wexler (29 Jul) |
| Agents | 500+ pre-built; Agent Builder; 25,000+ customer-built; 700,000+ agent tasks per day | Agentic OS (aOS), Workflows, Monitors, Portal |
| Own model | Harvey Tenet, post-trained open-weight, research preview 20 Aug 2026 | None announced |
| Reported pricing | ~$1,000–1,200 per seat per month, 20-seat minimum, plus consumption | From ~$3,000 per user per year, 10-seat minimum |
Sources: LawSites on Harvey's round; TechCrunch on Legora's extension; ILTA 2026 survey via LawSites. The Harvey review and Legora review cover each product on its own.
How the rivalry got here
The valuation ladder tells the story better than any product sheet.
- February 2025: Harvey valued at $3bn. Legora is a 40-person Stockholm startup.
- June 2025: Harvey $5bn Series E, with RELX (LexisNexis's parent) investing.
- September 2025: Linklaters rolls Legora out across 30 offices.
- October 2025: Legora $150m Series C at $1.8bn; Harvey $150m at $8bn, the same week.
- December 2025 to January 2026: White & Case picks Legora for 43 offices; CMS extends Harvey to 7,000-plus lawyers in 21 member firms across 50-plus countries.
- 3 February 2026: Anthropic's Claude legal plugin knocks 18% off Thomson Reuters and 14% off RELX in a day. Within three weeks Harvey announces an Anthropic integration.
- 10 March 2026: Legora $550m Series D at $5.55bn (Accel). 25 March: Harvey $200m at $11bn (GIC, Sequoia).
- 29–30 April 2026: Legora extension to $5.6bn; Slaughter and May goes firm-wide with Harvey.
- 7 May 2026: Legora launches aOS, its "agentic operating system". 5 May: Harvey announces 500-plus agents and Agent Builder early access.
- 14 May 2026: Baker McKenzie rolls Legora out globally. 16 June: Harvey–Microsoft Copilot integration.
- 18–20 August 2026: Harvey II (Memory, matter-aware Spaces, context-inheriting agents) and the Tenet model.
- 9 September 2026: Harvey $550m at $15.5bn. 16 September: Salesforce rolls Legora out to its global legal department. 17 September: Legora ships a ChatGPT Enterprise plugin.
Artificial Lawyer's read on the race, published the day of Harvey's round: "both Harvey and Legora are on the road to becoming giants", and with the new cash "we can expect more M&A" (Artificial Lawyer, 9 Sep 2026). The two have also fought a marketing war: Harvey hired Gabriel Macht of "Suits" for a TV advert; Legora hired Jude Law.
Product architecture: Workflow Agents against an Agentic OS
Harvey's stack is a set of named surfaces (Assistant, Vault, Knowledge, Workflow Agents, Agent Builder, Contract Intelligence, Word and Outlook add-ins) joined since Harvey II by Memory and matter-aware Spaces, where agents inherit the context of the matter they run in. Harvey defines an agent as something that can "plan, adapt, and meaningfully interact with humans" and a workflow as one or more agents producing "a meaningful, specific work product". The company reported 400,000-plus agentic queries a day in March 2026 and 700,000-plus agent tasks a day by June, and says its agents "flag moments requiring user confirmation before proceeding". The May 2026 brief added agentic playbook review, guest accounts and SCIM provisioning; a 23 July integration with Intapp enforces ethical walls.
Legora's pitch since 7 May 2026 is aOS: an operating layer, three years in development, in which overnight counterparty redlines are "reviewed, flagged and responded to by the time lawyers arrive". Around it sit Tabular Review (Addleshaw Goddard cites analysis of up to 100,000 documents in parallel), Workflows, Monitors, a client Portal and Word and Outlook add-ins. Junestrand's line at launch: "The ceiling of what's possible has always been limited by human capacity."
The honest difference is depth versus openness. Harvey has more pre-built, practice-specific agents and a larger catalogue of customer-built ones. Legora has been quicker to expose a general-purpose agent layer and to plug into other ecosystems, including the ChatGPT Enterprise plugin. For a description of what an agent is and where the failure modes sit, see what AI agents are.
Models: the same brains, one extra
Neither company is a model company, although Harvey is now edging that way. Harvey routes across OpenAI, Anthropic, Google and Mistral; Claude Opus 5 went live in Harvey on 24 July 2026 and Fable 5.1 on 1 September, and GPT-5.6 Sol scored 92.7% on Harvey's own BigLaw Bench in July. Harvey Tenet, a post-trained open-weight model in research preview since 20 August 2026, is the first vendor-specific legal model at this scale; Legora has announced nothing comparable and, on the evidence of its ChatGPT plugin, prefers to sit on top of the labs.
The benchmark Harvey published in May 2026, LAB, is open source (1,200-plus tasks, 24 practice areas, 75,000-plus rubric criteria) and shipped without a leaderboard. On its strict all-criteria-pass metric the best frontier model manages 13.3%. Bob Ambrogi's caution applies to both vendors: it is "a benchmark built by a market participant". The benchmarks guide explains how to read it.
Pricing, minimums and the direction of travel
No price list exists for either. The estimates that circulate in secondary sources are Harvey at roughly $1,000–1,200 per seat per month with a 20-seat minimum plus consumption elements, and Legora from about $3,000 per user per year with a 10-seat minimum. Treat both as opening positions. Reports in September 2026 point to token-based components in Harvey contracts, and Artificial Lawyer notes that "in a world where usage-based (i.e. token) pricing is becoming more relevant to some companies, then such companies really, really want you to stay where you are".
The arithmetic for a mid-sized firm: 30 seats of Harvey at the reported rate is $360,000–432,000 a year before consumption; 30 seats of Legora at the reported entry rate is $90,000. That gap narrows sharply in enterprise negotiations and widens again once usage is billed. The Harvey pricing question and the legal AI pricing guide set both against CoCounsel, Clio Work at $199 and the $25 chat tools.
Geography and integrations
Harvey is American in customer base and cadence: 84% of firms with 700-plus lawyers in the ILTA survey, design-partner status in OpenAI's Astra for Law plugins, a Microsoft Copilot integration since 16 June 2026 and corporate clients including HSBC, HubSpot, P&G, PwC, T-Mobile and Merck. Its EMEA footprint is real (CMS, Slaughter and May, Hengeler Mueller, Norton Rose Fulbright, GSK Stockmann, Schoenherr in RSGI's study), but the centre of gravity is the US.
Legora is European by origin and increasingly global by client list: Linklaters, White & Case, Baker McKenzie, Cleary, Bird & Bird, Simmons & Simmons, Addleshaw Goddard, Deloitte Legal UK, Barclays, Erste Group in Vienna, and now Salesforce. Its acquisitions in 2026 were a Canadian firm (Walter AI), a research engine (Qura) and a litigation fact-analysis tool (Wexler). For a European firm the practical advantages are proximity, a vendor that already answers GDPR questions in EU terms, and reference customers a train ride away.
Both connect to the same document stores. The iManage MCP Server (14 May 2026) works with Claude, ChatGPT, Copilot, Harvey, Legora and custom agents; NetDocuments' MCP runs under the user's own credentials. Both are also partner plugins in Astra for Law, and Harvey is a connector in Claude for Legal. The MCP setup guide explains what that permission model means in practice.
What the sceptics say
Three lines of doubt deserve airtime before you sign.
Valuation. Harvey trades at roughly 39 times ARR; Legora is chasing $10bn on roughly $150m of revenue. Robin AI, once the UK's best-funded legal AI startup, failed to close a $50m round, cut about 50 roles and sold itself to Microsoft as an acqui-hire in January 2026. The funded-company count in legal tech fell 27% in 2025 while total funding rose 22%; money is concentrating in a few names, which is good for those names and bad for negotiating power.
Centrality. Richard Tromans frames the market as a "battle for centrality": OpenAI, Anthropic, Google, Microsoft, Harvey, Legora, Thomson Reuters and LexisNexis all want to be the place "where your legal work primarily lives". Harvey's CEO is reported to have conceded that "long term we would end up competing with the model companies". Law.com asked in July 2026 "Who is not using Harvey or Legora, and why?", and Latham & Watkins' answer was to buy its own GPUs while keeping both subscriptions.
Evidence. Independent evidence is thin. RSGI's study of 40 Harvey organisations (November 2025) found 68% saw measurable benefit within three months and power users saved about 37 hours a month; CMS reported 93% of pilot users seeing productivity gains and up to 117.9 hours saved per lawyer per year. Both are customer-reported. Neither vendor has submitted to a public head-to-head since Vals' February 2025 report, where Harvey scored 94.8% on document Q&A but lawyers still beat it at redlining (79.7% against 65.0%).
Who should pick which
| If you are | Lean towards | Because |
|---|---|---|
| A US firm of 100+ lawyers with corporate clients on Harvey | Harvey | Client alignment, 84% peer adoption, Copilot integration |
| A European firm of 20–200 lawyers | Legora | 10-seat minimum, EU vendor, Tabular Review, local references |
| A firm whose work is diligence-heavy | Either; test Tabular Review against Harvey's Vault and diligence agents | GSK Stockmann reports 15–20% savings on standard diligence and up to 75% on unstructured data rooms with Harvey; Addleshaw Goddard cites 100,000-document Tabular Review with Legora |
| An in-house team inside a Salesforce, Barclays or Erste-style corporate | Legora | Growing corporate references; Portal for business users |
| An in-house team at a bank or pharma group already on Harvey via outside counsel | Harvey | HSBC, Merck, P&G precedent; shared workspaces with firms |
| A firm under 10 lawyers | Neither yet | Both minimums exclude you; see Harvey for small firms |
| A firm wanting to build its own agents on top | Harvey Agent Builder or Legora aOS, scored in a pilot | 25,000+ customer-built agents on Harvey; Simmons & Simmons runs "hundreds of agents" alongside Legora |
A pilot scorecard before you sign
Both vendors will offer a pilot. Run it as a test, not a demo. The six-week pilot playbook has the full design; the minimum scorecard is below.
- Pick three tasks per practice group (for example: NDA redline against your playbook, a 200-document diligence grid, a memo on a recurring question) and freeze the inputs before the pilot starts.
- Score each task on accuracy against a partner-marked answer, minutes saved against a timed baseline, and the number of corrections needed before it could go to a client.
- Record every hallucinated citation, invented clause or silently dropped instruction. One is information; three is a pattern.
- Test the DMS connector under a real ethical wall and confirm the audit log shows who ran what.
- Get the price schedule in writing, including consumption, overage, seat-minimum and exit terms, and run it through the vendor due-diligence questions.
- Ask both vendors what happens to your prompts, outputs and custom agents on termination.
Session 4 of the AI Lab for Lawyers course walks through Harvey and Legora side by side so that participants can score them on their own work rather than on a vendor's demo. That is the only comparison that counts.
Frequently asked questions
Is Harvey or Legora better?
Neither is better in the abstract. Harvey is larger, US-centred and deeper in pre-built agents, with 3,000-plus customers and 80 percent of the top 100 law firms as of September 2026. Legora is European, cheaper to start with, and known for Tabular Review across up to 100,000 documents. Pick by geography, seat count, the practice groups that will use it most, and the result of a scored pilot.
How much do Harvey and Legora cost?
Neither publishes prices. Secondary sources put Harvey at about $1,000 to $1,200 per seat per month with a 20-seat minimum plus consumption elements, and Legora from about $3,000 per user per year with a 10-seat minimum. Enterprise deals vary widely, and both vendors have been reported to move towards usage-based components, so ask for a written price schedule before a pilot.
Which law firms use Harvey and which use Legora?
Harvey lists 80 percent of the top 100 law firms, including Slaughter and May firm-wide, CMS across 7,000 lawyers in 21 member firms, DLA Piper with 5,000 licences, and in-house teams such as HSBC and PwC. Legora names Linklaters across 30 offices, White & Case across 43, Baker McKenzie, Cleary, Bird & Bird, Simmons & Simmons, and Salesforce's global legal department. Latham & Watkins uses both.
Do Harvey and Legora use Claude or GPT?
Both are multi-model. Harvey routes across OpenAI, Anthropic, Google and Mistral models; Claude Opus 5 went live in Harvey on 24 July 2026 and Fable 5.1 on 1 September 2026, and Harvey added its own open-weight model, Tenet, in August 2026. Legora also runs frontier models from several labs and launched a ChatGPT Enterprise plugin on 17 September 2026.
Put this into practice
AI Lab for Lawyers: a live, hands-on course for practising lawyers
4 live sessions (8 hours), recordings included, taught by Dr. Niklas Schmidt of Wolf Theiss. Rated 4.7/5 from 381 ratings; alumni from Norton Rose Fulbright, Paul Hastings, Pillsbury and Taylor Wessing.
Related reading
Tools
Harvey AI in 2026: What It Does, What It Costs and Who Uses It
Harvey AI reviewed as of September 2026. The $15.5bn valuation, Harvey II, Tenet, 500-plus agents, who uses it, estimated seat prices and its own benchmark.
9 min read
Tools
Legora Review: The Agentic OS Challenging Harvey
Legora reviewed as of September 2026. The $5.6bn Stockholm challenger to Harvey: aOS, Tabular Review, Wexler litigation facts, pricing, clients to Salesforce.
7 min read
Questions
How Much Does Harvey AI Cost? Pricing, Minimums and Alternatives
Harvey AI costs an estimated $1,000–1,200 per seat per month with a 20-seat minimum as of September 2026. What you get, how to negotiate and cheaper options.
6 min read
Guides
What Legal AI Really Costs in 2026: Harvey, Legora, CoCounsel, Claude and ChatGPT
Legal AI cost as of September 2026, from €39 to $1,200 per seat per month, with Harvey, Legora, CoCounsel, Claude and ChatGPT prices and worked budgets for 5 and 50 lawyers.
10 min read
Playbooks
Run a Six-Week AI Pilot in Your Firm: Design, Metrics and the Go/No-Go
Run a law firm AI pilot in six weeks: three use cases, ten users, a 20-task evaluation set, baseline metrics, security sign-off and a go/no-go scorecard.
11 min read
Regulation & Ethics
AI Vendor Due Diligence for Law Firms: The Questions to Ask Before You Sign
AI vendor due diligence for law firms starts with the subprocessor chain, then training and retention terms, ZDR eligibility, data residency, audit logs and exit rights.
11 min read