Lexis+ AI Review: vs Westlaw CoCounsel, Scored
Lexis+ AI review vs Westlaw CoCounsel, scored on an independent 50-point benchmark. Lexis wins research (41), CoCounsel wins drafting. Pricing and the gap.
Search for a Lexis+ AI review and you get two kinds of page: LexisNexis marketing, and affiliate directories that score the product on screenshots. Almost none of them put it in a ring with Westlaw CoCounsel and a published rubric.
This review does. We test LexisNexis +AI, CoCounsel, and 17 other models and platforms across 11 legal task categories on a public 50-point benchmark, and the full table is on our compare page for anyone to check.
Full disclosure up front: HAQQ is a competitor. We build legal AI MENA-first, with native Arabic, for consumers and small firms as well as professionals. We are not neutral. So this review keeps opinions cheap and numbers expensive: every figure is either from our published benchmark or from a named external source you can click.
Key facts about Lexis+ AI
- In February 2026, LexisNexis launched Lexis+ with Protégé, which it says "fully replaces Lexis+ AI" as its flagship legal AI platform, according to LawNext (2026).
- It is grounded in Shepard's Citations verification and a living legal knowledge graph of more than 200 billion interconnected documents, with over 4 million added daily, according to LawNext and LexisNexis (2026).
- Protégé routes through third-party frontier models from Anthropic, Google, and OpenAI behind a single conversational prompt box, according to LawNext (2026).
- Scored 37/50 on the generic evaluation of the independent benchmark on our compare page, versus 49/50 for HAQQ. Average across all 11 categories: 33.2 vs 47.5.
- Strongest category by far: legal research at 41/50, where it ranked fifth overall and first among every legal-vertical platform we tested, ahead of CoCounsel (36) and Harvey (37).
- Weakest categories: drafting. It scored 30 to 34 on agreements and law explanation, below CoCounsel on nine of eleven categories.
- A 2024 Stanford RegLab study, peer-reviewed in 2025, found Lexis+ AI hallucinated on more than 17% of queries despite LexisNexis's "hallucination-free" marketing, according to Stanford HAI.
- US-and-English-first. No native-Arabic, no consumer tier, no self-serve SMB plan. Reported pricing runs roughly $250 to $475 per user per month.
What is Lexis+ AI (now Lexis+ with Protégé)?
Lexis+ AI is LexisNexis's generative-AI layer on top of its legal research database. In February 2026 LexisNexis rebuilt it into Lexis+ with Protégé, a platform it says "fully replaces Lexis+ AI," with one conversational prompt box that spans research, drafting, summarization, and document analysis, according to LawNext (2026).
The thing that makes it Lexis and not a chatbot is the content underneath. Protégé sits on a living legal knowledge graph of more than 200 billion interconnected documents, more than 4 million added daily, and grounds answers in LexisNexis primary law, secondary sources, and Shepard's Citations, according to LawNext and LexisNexis (2026). The model itself is borrowed: Protégé routes through Anthropic, Google, and OpenAI frontier models. The moat is the data and the citation graph, not the weights.
That is the whole Lexis pitch in one sentence: a frontier model you already trust, pointed at a legal corpus you cannot get anywhere else, with every citation checked against Shepard's. For a US litigator, that corpus is the product.
Lexis+ AI's citation and grounding approach
Grounding is where Lexis differentiates, so it deserves its own section. Instead of asking a model to recall case law from training, Protégé retrieves from the LexisNexis corpus and runs Shepard's Verify Trust Markers, which scan both AI-generated and attorney-drafted text, find every citation, and check it against authoritative sources. The goal is a workflow where the model drafts and Shepard's audits.
Here is the honest part. Grounding reduces hallucination; it does not eliminate it. A 2024 Stanford RegLab study, later peer-reviewed in the Journal of Empirical Legal Studies in 2025, tested these tools head-to-head and found Lexis+ AI produced incorrect information on more than 17% of queries, while Westlaw's AI-Assisted Research hallucinated on more than 34%, according to Stanford HAI. LexisNexis had publicly marketed "hallucination-free linked legal citations." The researchers' verdict was blunt: "Providers' claims are overstated."
Lexis+ AI benchmark scores: 37 generic, 41 on research
The independent benchmark we publish on our compare page scores 19 models and platforms on a 50-point generic rubric covering Sharia, statute, forum, clause, risk, hallucination, formatting, brevity, partner-readiness, and source linking, plus ten further categories scoring specific deliverables.
LexisNexis +AI scored 37/50 on the generic evaluation. The interesting story is not the headline number, it is the shape: Lexis is built like a research tool, and the category scores prove it.
- Legal research is its peak: 41/50. That is fifth overall, behind only HAQQ (48) and three raw frontier models, and it is the best score of any legal-vertical platform on this task. It beat CoCounsel (36), Harvey (37), and Legora (31). The content moat is real.
- Drafting is its valley. It scored 32 on contract drafting, 30 on law explanation, 31 on license and consultancy agreements, and 32 on employment, shareholder, and commercial agreements. On nine of eleven categories, CoCounsel outdrafted it.
- It never finishes in the top three on anything except by content. On research it is the strongest vertical; everywhere else it sits mid-table, behind raw Claude and ChatGPT on answer quality.
Here is Lexis+ AI against Westlaw's CoCounsel and HAQQ across all 11 categories:
| Task category | Lexis+ AI /50 | CoCounsel /50 | HAQQ /50 |
|---|---|---|---|
| Generic legal evaluation | 37 | 37 | 49 |
| Contract drafting | 32 | 36 | 47 |
| Legal research | 41 | 36 | 48 |
| Law explanation | 30 | 31 | 46 |
| Employment agreement | 32 | 37 | 48 |
| Professional memorandum | 34 | 39 | 46 |
| License agreement | 31 | 35 | 47 |
| Shareholder agreement | 32 | 37 | 48 |
| Consultancy agreement | 31 | 36 | 47 |
| Commercial agreement | 32 | 37 | 48 |
| NDA drafting | 33 | 37 | 49 |
| Average | 33.2 | 36.2 | 47.5 |
Lexis+ AI vs Westlaw CoCounsel: the incumbent head-to-head
This is the comparison most buyers actually want, because for a US firm the choice often comes down to which research empire you already pay: LexisNexis or Thomson Reuters. Both wrapped a frontier model around a proprietary corpus. They diverge on what they are good at.
Research: Lexis wins
On our benchmark, Lexis+ AI scored 41 on legal research to CoCounsel's 36, a five-point gap and the single clearest result in the matchup. The Stanford RegLab study points the same direction on reliability: Lexis hallucinated on roughly 17% of queries, while Westlaw's AI-Assisted Research, the research engine behind the CoCounsel ecosystem, hallucinated on more than 34%, according to Stanford HAI. If your primary job is finding and validating authority, Lexis is the stronger of the two incumbents.
Drafting: CoCounsel wins
Flip to drafting and the result flips. CoCounsel outscored Lexis on nine of eleven categories: 39 to 34 on professional memoranda, 37 to 32 on employment and shareholder agreements, 36 to 32 on contract drafting. CoCounsel's average across all eleven categories is 36.2 to Lexis's 33.2. Thomson Reuters built CoCounsel around drafting and review workflows, and the scores show it.
Pricing: both are expensive, and neither is transparent
Neither lists clean public pricing, so every figure here is a third-party 2026 estimate, not a quoted rate. Legal-tech pricing analyses in 2026 put Lexis+ with AI at roughly $250 to $475 per user per month for mid-size firms. CoCounsel is bundled with Westlaw and not sold standalone: estimates put the CoCounsel add-on around $100 to $200 per user per month on top of a Westlaw subscription of $200 to $400-plus, so the real all-in cost lands closer to $300 to $600 per user per month, per AI Vortex and TheLegalPrompts (2026). Both are firm-budget products. Verify any number against an actual quote before you rely on it.
The honest summary of Lexis+ AI vs CoCounsel: buy Lexis if research is the job, buy CoCounsel if drafting is the job, and budget for a frontier-model wrapper either way.
Where Lexis+ AI falls short
Drafting is mid-table
The benchmark is direct: across the nine drafting categories Lexis trails CoCounsel and, on answer quality, trails raw Claude and ChatGPT too. You buy Lexis for its corpus and its citations, not for the prose. If your week is mostly agreements and memos, you are buying the wrong incumbent.
Grounding is not a guarantee
The 17% Stanford hallucination figure is the most important caveat in any Lexis review. "Grounded in Shepard's" is a real advantage over an ungrounded chatbot, but it is not the "hallucination-free" the marketing once claimed. The verification duty still sits with the lawyer.
Language and market: US-and-English-first
The LexisNexis corpus is deepest where LexisNexis has always been deepest: US and common-law content in English. There is no native-Arabic interface, no consumer tier, and no self-serve SMB plan. A solo lawyer in Cairo, a three-person firm in Dubai, a founder who needs one shareholder agreement checked under UAE law: none of them are in the Lexis+ AI market, the same gap we mapped across the region in our MENA legal tech coverage.
When Lexis+ AI is the right choice
Fair is fair. Lexis+ with Protégé is a defensible, probably correct choice if most of these are true:
- Your practice is research-heavy US or common-law work, and authority validation is the core task.
- You already pay for LexisNexis and want AI on top of a corpus you trust.
- Shepard's citation validation is part of your existing quality process.
- You have a firm budget and a procurement process, not a credit card.
- Your work is in English, in jurisdictions where the LexisNexis corpus is deep.
If that is you, Lexis+ AI belongs on your shortlist, and on research it is the stronger of the two incumbents. If your work is drafting-heavy, weigh CoCounsel first.
Lexis+ AI and CoCounsel alternatives by use case
| Use case | Strongest fit | Why |
|---|---|---|
| US research-heavy practice | Lexis+ AI | 41/50 on research, best vertical platform, Shepard's grounding |
| US drafting and review | CoCounsel | Westlaw content + best incumbent drafting scores (36.2 avg) |
| MENA, Arabic, civil-law jurisdictions | HAQQ Legal AI | Native Arabic with RTL, 80+ countries, 49/50 generic, 48/50 research |
| Solo lawyers, SMBs, consumers | HAQQ Legal AI | Self-serve, free tier, $33 to $100/mo |
| Enterprise English drafting at scale | Harvey or Legora | Largest deployed footprints; enterprise-only |
| Build-your-own on a frontier model | Claude | 45/50 generic at API prices, but you own verification |
HAQQ Legal AI: for the market both incumbents skip
The obvious caveat: this is us. HAQQ is built MENA-first with native Arabic and RTL, covers 80+ countries, and serves consumers and small firms as well as professionals, with a free tier and paid plans from $33 to $100 per month. It scored 49/50 on the generic evaluation and 48/50 on legal research, ranking first in all 11 categories, and unlike most of the market we disclose our engine, Justinian. If your work touches UAE, Saudi, Egyptian, Lebanese, or Qatari law, or you are simply not a firm-budget US buyer, this is the gap both Lexis and CoCounsel leave open. You can test it free in minutes.
CoCounsel: the Westlaw answer
CoCounsel from Thomson Reuters is Lexis's direct rival, bundled with Westlaw and built around drafting and review. It scored 37 generic and won nine of eleven drafting categories against Lexis, but trailed it on research. If you are choosing between the two incumbents, read our CoCounsel review and alternatives.
Raw frontier models: the DIY route
Protégé already routes through Anthropic, Google, and OpenAI models, so you can ask why not use them directly. The answer is the corpus and the verification layer. Claude Fable 5 scored 45/50 on our generic evaluation, but in our separate 300-task frontier benchmark, 24% of all frontier-model answers cited or applied law that did not say what the model claimed. The grounding and citation layer is the thing Lexis sells, and the thing you would have to build yourself.
FAQ
What is Lexis+ AI?
Lexis+ AI is LexisNexis's generative-AI layer over its legal research database. In February 2026 LexisNexis relaunched it as Lexis+ with Protégé, which it says "fully replaces Lexis+ AI," combining research, drafting, and analysis in one conversational prompt box grounded in Shepard's Citations and a 200-billion-document knowledge graph, according to LawNext (2026).
Is Lexis+ AI better than Westlaw CoCounsel?
It depends on the task. On our 50-point benchmark, Lexis+ AI won legal research (41 vs 36) while CoCounsel won nine of eleven drafting categories and the overall average (36.2 vs 33.2). Buy Lexis for research, CoCounsel for drafting. Both are US-and-English-first firm-budget products.
How much does Lexis+ AI cost?
LexisNexis does not publish clean public pricing. Third-party 2026 analyses estimate roughly $250 to $475 per user per month for mid-size firms. CoCounsel is bundled with Westlaw and estimated higher all-in, around $300 to $600 per user per month, per AI Vortex and TheLegalPrompts (2026). Verify against a real quote.
Does Lexis+ AI hallucinate?
Yes, less than an ungrounded chatbot but not zero. A 2024 Stanford RegLab study, peer-reviewed in 2025, found Lexis+ AI produced incorrect information on more than 17% of queries, and Westlaw's AI-Assisted Research on more than 34%, despite LexisNexis's "hallucination-free" marketing, according to Stanford HAI. Every cited source still needs a human to read it.
What did Lexis+ AI score on an independent benchmark?
On the 50-point benchmark on our compare page, LexisNexis +AI scored 37/50 on the generic evaluation and 41/50 on legal research, where it was the strongest legal-vertical platform we tested. It averaged 33.2 across all 11 categories, versus 37 generic and 36.2 average for CoCounsel and 49 and 47.5 for HAQQ.
Does Lexis+ AI work in Arabic or for MENA law?
Not natively. Lexis+ AI is US-and-English-first, with no native-Arabic interface, no consumer tier, and a corpus deepest in US and common-law content. For Arabic-language, civil-law, MENA work, a MENA-first tool like HAQQ, which scored 49/50 generic and covers 80+ countries with native Arabic and RTL, is the better fit.
- HAQQ vs other legal AI: the full benchmark
- Best AI for legal work: 300-task frontier benchmark
- CoCounsel review and alternatives
- When AI lies to the court
- Best legal AI tools 2026
- MENA legal tech market
- LawNext: LexisNexis launches Lexis+ with Protégé, replacing Lexis+ AI
- Stanford HAI: AI on Trial, legal models hallucinate in 1 of 6 queries
- Stanford RegLab: Hallucination-Free? Assessing leading AI legal research tools
- AI Vortex: CoCounsel pricing 2026
- TheLegalPrompts: Legal AI pricing 2026, Harvey vs CoCounsel vs Clio
FAQ
What is Lexis+ AI?
Lexis+ AI is LexisNexis's generative-AI layer over its legal research database. In February 2026 LexisNexis relaunched it as Lexis+ with Protégé, which it says "fully replaces Lexis+ AI," combining research, drafting, and analysis in one conversational prompt box grounded in Shepard's Citations and a 200-billion-document knowledge graph, according to LawNext (2026).
Is Lexis+ AI better than Westlaw CoCounsel?
It depends on the task. On our 50-point benchmark, Lexis+ AI won legal research (41 vs 36) while CoCounsel won nine of eleven drafting categories and the overall average (36.2 vs 33.2). Buy Lexis for research, CoCounsel for drafting. Both are US-and-English-first firm-budget products.
How much does Lexis+ AI cost?
LexisNexis does not publish clean public pricing. Third-party 2026 analyses estimate roughly $250 to $475 per user per month for mid-size firms. CoCounsel is bundled with Westlaw and estimated higher all-in, around $300 to $600 per user per month, per AI Vortex and TheLegalPrompts (2026). Verify against a real quote.
Does Lexis+ AI hallucinate?
Yes, less than an ungrounded chatbot but not zero. A 2024 Stanford RegLab study, peer-reviewed in 2025, found Lexis+ AI produced incorrect information on more than 17% of queries, and Westlaw's AI-Assisted Research on more than 34%, despite LexisNexis's "hallucination-free" marketing, according to Stanford HAI. Every cited source still needs a human to read it.
What did Lexis+ AI score on an independent benchmark?
On the 50-point benchmark on our compare page, LexisNexis +AI scored 37/50 on the generic evaluation and 41/50 on legal research, where it was the strongest legal-vertical platform we tested. It averaged 33.2 across all 11 categories, versus 37 generic and 36.2 average for CoCounsel and 49 and 47.5 for HAQQ.
Does Lexis+ AI use third-party AI models?
Yes. Lexis+ with Protégé routes through frontier models from Anthropic, Google, and OpenAI behind a single conversational prompt box, according to LawNext (2026). The differentiation is not the model, it is the LexisNexis corpus and the Shepard's Citations verification layer wrapped around it.
Does Lexis+ AI work in Arabic or for MENA law?
Not natively. Lexis+ AI is US-and-English-first, with no native-Arabic interface, no consumer tier, and a corpus deepest in US and common-law content. For Arabic-language, civil-law, MENA work, a MENA-first tool like HAQQ, which scored 49/50 generic and covers 80+ countries with native Arabic and RTL, is the better fit.