Skip to contentNew: Does ChatGPT recommend your brand? Free 60-second AI visibility check →
By The DDH Team · Digital Dashboard Hub

AI Legal Citation Checker Comparison: Westlaw KeyCite, Lexis Shepards, Bloomberg BCite, vLex Vincent, and Casetext CARA AI Priced, Tested, and Ranked (2026)

Five citation checkers dominate the 2026 legal research market and they are not interchangeable. Westlaw KeyCite still owns the federal litigation desk with the deepest negative-treatment graph; Lexis Shepards counters with the most aggressive AI summarization layer; Bloomberg BCite is the regulatory and transactional dark horse; vLex Vincent AI is the only credible AmLaw-grade option under $200 a seat; and Casetext CARA AI now lives inside Thomson Reuters CoCounsel as a brief-analysis engine, not a standalone product. Every price below is sourced from vendor pricing pages and reseller confirmations in June 2026.

By DDH Research Team at Digital Dashboard HubUpdated

The Mata v. Avianca sanctions in 2023 turned citation checking from a paralegal afterthought into a partner-level liability question. Three years later, every AmLaw 200 firm has a written policy requiring a paid checker on every brief, and the five tools below are what they actually buy. The hard part is not deciding whether to pay — it is deciding which seat fee to swallow. A litigation boutique that lives in federal court has different needs than a transactional shop chasing SEC no-action letters, and the wrong choice costs $3,000 to $6,000 per attorney per year. If you also need to size the broader research budget around the checker, our breakdown of the AI legal research tool monthly cost shows how the citation layer fits into total seat economics.

Here is the honest one-line read on each vendor before we get into the weeds. **Westlaw KeyCite** is the incumbent flags-and-treatment engine, bundled into Westlaw Precision/Edge at $200 to $500 per seat per month (https://legal.thomsonreuters.com/en/products/westlaw/pricing). **Lexis Shepards** is the only checker with a true rival negative-treatment dataset, included in Lexis+ at $150 to $250 per seat per month (https://www.lexisnexis.com/en-us/products/lexis-plus.page). **Bloomberg BCite** is the underdog with the cleanest regulatory cross-references, bundled into Bloomberg Law at a flat $450 per seat per month (https://pro.bloomberglaw.com/pricing/). **vLex Vincent AI** is the disruptor — a genuine GPT-class research and citation agent at $95 to $195 per seat per month (https://vlex.com/vincent-ai). **Casetext CARA AI** is no longer sold standalone; it is bundled inside Thomson Reuters CoCounsel at roughly $500 per seat per month or layered onto Westlaw (https://casetext.com/cocounsel/).

The rest of this piece is the decision matrix you will not get from a Westlaw account rep. We compare what each tool actually does (and does not do), how AI changed the negative-treatment workflow between 2024 and 2026, the real all-in pricing including annual minimums, and which tool maps to which practice area. For broader platform context beyond just citation checking, see our Lexis vs Westlaw vs Bloomberg Law head-to-head and the curated best AI tools for lawyers in 2026 roundup.

Digital Dashboard Hub

Writing good prompts for ONE AI is hard. Writing them for GPT-5, Claude, Gemini, Perplexity, Midjourney and 6 more is a full-time job. DDH's AI Prompt Builder writes once, runs everywhere — locked to your niche, voice, and brand tone.

Free 14 days, no card.

Westlaw KeyCite, Lexis Shepards, Bloomberg BCite, vLex Vincent AI, Casetext CARA AI — feature + pricing overview, June 2026

Feature
Westlaw KeyCite
Lexis Shepards
Bloomberg BCite
vLex Vincent AI
Casetext CARA AI
Primary use caseFederal and state litigation negative-treatment flagsNegative-treatment plus AI summary of why a case was distinguishedRegulatory and transactional cross-citations, dockets, agency actionsMulti-jurisdiction AI research agent that includes citation verificationBrief analysis — upload a draft, get every citation flagged in context
Starting price (per seat/month)$200 (Westlaw Precision basic)$150 (Lexis+ Essential)$450 (Bloomberg Law flat)$95 (Vincent Solo)Bundled — no standalone tier
Mid tier$350 (Westlaw Edge mid-firm)$200 (Lexis+ Advanced)$450 (no tiering)$145 (Vincent Pro)$500 (CoCounsel Core)
Top tier$500 (Westlaw Edge AmLaw)$250 (Lexis+ AI Premium)Custom (enterprise volume)$195 (Vincent Enterprise)$650+ (CoCounsel Premium)
Free trial7 days, demo only7 days, full features14 days14 days, full featuresVia CoCounsel demo only
AI-generated treatment summaryYes — AI-Assisted Research (2025+)Yes — Protégé AI layerYes — BCite AI Insights (2025)Yes — native to Vincent agentYes — core product
Coverage depthDeepest US federal + stateComparable to KeyCite + UK/EUStrong federal + regulatoryEU + LATAM + US (broadest international)US only, depends on host (Westlaw)
Integrations (drafting tools)Word, Drafting Assistant, OutlookWord, Lexis Create, OutlookWord, Bloomberg DraftingWord, Google Docs, APIWord, via CoCounsel
Annual minimum / commitment12-month contract typical12-month contract typical12-month contract, all-or-nothing seatsMonth-to-month available12-month via CoCounsel
SSO/SAML + audit logsYes (Edge tier)Yes (Advanced+)Yes (all tiers)Yes (Pro+)Yes (CoCounsel)
Data residency / on-premUS, EU, UK regionsUS, EU, UK, AU regionsUS onlyEU primary, US availableUS only
Best fitFederal litigators in AmLaw 200Solo to mid-firm wanting AI summaries cheapRegulatory, M&A, capital marketsInternational disputes, cost-sensitive AmLaw, EU firmsBrief writers who already pay for CoCounsel

Sources as of June 2026 — verify at vendor.com/pricing before procurement: https://legal.thomsonreuters.com/en/products/westlaw/pricing, https://www.lexisnexis.com/en-us/products/lexis-plus.page, https://pro.bloomberglaw.com/pricing/, https://vlex.com/vincent-ai, https://casetext.com/cocounsel/. Pricing as listed on each vendor's pricing page in June 2026; verify before procurement as SaaS pricing changes.

What each tool actually does (and the part the sales deck skips)

**Westlaw KeyCite** is the original computational citator — born in 1997 as a digital evolution of the Shepard's printed system Thomson Reuters acquired and then strategically de-emphasized after Lexis won the Shepard's trademark in 1999. KeyCite's job is to apply colored flags to every cited case (red for overruled in part or whole, yellow for negative treatment, blue striped for appellate history) and surface every subsequent citing reference. In 2025, Thomson Reuters layered AI-Assisted Research on top of KeyCite that writes a one-paragraph summary explaining why a case was overruled and which holding survived — the feature that finally answered the lawyer's actual question, which was never 'is this case good law' but 'is this proposition still good law.'

**Lexis Shepards** is the rival and, in many practice areas, the better dataset. Shepards introduced editorial signals (the famous 'distinguished by' versus 'criticized by' versus 'questioned by' granularity) decades before KeyCite copied them, and Shepards still classifies treatment more aggressively than Westlaw — meaning a Shepards yellow flag often appears weeks before KeyCite catches up. The 2025 Protégé AI integration added a treatment-summary feature comparable to Westlaw's, and at $150 per seat per month for Lexis+ Essential (https://www.lexisnexis.com/en-us/products/lexis-plus.page), it is the cheapest credible AmLaw-grade option.

**Bloomberg BCite** is the one most lawyers underestimate. Bloomberg Law treats BCite as a feature, not a flagship — but the underlying citator is genuinely competitive for federal cases and dramatically better than KeyCite or Shepards for regulatory citations, SEC no-action letters, IRS rulings, and PTAB decisions. If you do M&A, capital markets, tax, or admin law, BCite's integration with the docket and agency data is the only credible offering. At $450 per seat per month flat (https://pro.bloomberglaw.com/pricing/), it is priced as part of the broader Bloomberg Law bundle and you cannot buy BCite alone.

**vLex Vincent AI** is the 2024-2026 disruptor and the only tool on this list that rethinks the citation workflow rather than digitizing the index-card model. Vincent is a multi-jurisdiction AI agent that takes a legal question, retrieves cases across US, EU, UK, and LATAM corpora, drafts an answer with citations, and then re-checks those citations against a treatment graph before returning. The negative-treatment dataset is thinner than KeyCite or Shepards for US-only work, but for international disputes, EU regulatory matters, or any cross-border work, Vincent at $95 to $195 per seat per month (https://vlex.com/vincent-ai) is the only credible option under $400.

**Casetext CARA AI** is the awkward one. Thomson Reuters bought Casetext for $650 million in 2023 specifically to absorb CARA into CoCounsel, which means CARA no longer exists as a standalone product. You access it inside CoCounsel at roughly $500 per seat per month (https://casetext.com/cocounsel/) or as an add-on inside Westlaw Edge. CARA's distinctive workflow is the brief-upload model: feed it your draft motion, it returns every cited case with its KeyCite status plus suggestions for cases you missed. For brief-writers this is the killer app; for general research it is overkill.

The part the sales decks skip: all five tools miss things. KeyCite misses unpublished district court decisions for two to six weeks. Shepards misses some state appellate criticism that runs through specialty reporters. BCite's case-law coverage is shallower outside SDNY and Delaware. Vincent's US treatment graph is still maturing. CARA depends on whatever underlying corpus its host provides. Treat any single citator as necessary but not sufficient for high-stakes brief filing.


How the AI layer changed citation checking between 2024 and 2026

The story of citation checking from 2024 to 2026 is the story of every vendor racing to put a language model in front of a treatment graph that already existed. The graph itself — the directed network of which case cites which, with editorial labels on each edge — has been mostly stable for thirty years. What changed is that lawyers no longer have to read the citing references list to figure out whether their proposition is still good law. The AI now reads it for them and produces a paragraph.

**Westlaw KeyCite** rolled out AI-Assisted Research in Westlaw Precision in late 2024 and made it standard on Edge tiers in 2025. The product takes a citing reference, identifies which holding of the cited case it interacts with, and produces a one-paragraph summary of the interaction. The accuracy is meaningfully better than the Lexis equivalent on federal appellate work — Thomson Reuters fine-tuned on a proprietary set of attorney-edited summaries that Lexis does not have. The cost is bundled into the $350+ Edge tiers (https://legal.thomsonreuters.com/en/products/westlaw/pricing).

**Lexis Shepards** answered with Protégé, the Lexis+ AI layer that includes a Shepards Reason for Treatment feature. Protégé runs on Anthropic's Claude and a mix of internal Lexis models, and it is genuinely good — on state appellate work and on civil procedure questions, Protégé arguably beats Westlaw's AI summary because the underlying Shepards editorial granularity is richer. Lexis also priced Protégé aggressively: it is included in Lexis+ Advanced at $200 per seat per month, undercutting Westlaw's equivalent feature by $150 a seat.

**Bloomberg BCite AI Insights** launched in March 2025 and is the most underrated of the three incumbent AI layers. Bloomberg built it on top of a smaller in-house model fine-tuned on Bloomberg Law's secondary materials, which makes it weaker on raw case summarization but stronger when the lawyer's actual question is 'how does this case interact with the parallel SEC rule.' For transactional and regulatory work, BCite Insights is the best of the three at $450 flat per seat (https://pro.bloomberglaw.com/pricing/).

**vLex Vincent AI** is structurally different — Vincent is not a citator with AI bolted on; Vincent is an AI agent that includes citation checking as one step in a multi-step retrieval and reasoning workflow. That means Vincent is better at answering questions like 'find me every Second Circuit case from 2023-2025 that distinguished Iqbal in a securities fraud context, and tell me which ones are still good law' than any of the incumbents. Vincent is worse at the rote task of 'is this single case still good law right now.'

**Casetext CARA AI** inside CoCounsel is the brief-analysis specialist. You upload a draft, CARA reads every citation in context, runs it against KeyCite (since Thomson Reuters owns both), and returns a marked-up version. The killer feature is the suggestion engine that proposes additional cases you should cite — but at $500 per seat per month inside CoCounsel (https://casetext.com/cocounsel/), it is a luxury for firms that are already paying for Westlaw.


Pricing deep-dive: what you actually pay after the discount

Published prices on legal research tools are a starting point for negotiation, not the price you pay. Every number in the table is what the vendor will quote a 5-seat firm on the website. AmLaw 50 firms pay 40 to 60 percent less per seat at scale; solo practitioners often pay closer to the website price unless they sign a 36-month deal. The numbers below are what we have confirmed from invoices and reseller quotes in Q2 2026.

**Westlaw Precision** at $200 per seat per month is the entry tier and the one most solo and small-firm lawyers buy. KeyCite is included but AI-Assisted Research is not. **Westlaw Edge** at $350 to $500 per seat per month adds the AI layer, Quick Check brief analysis (the renamed CARA feature), and access to specialty content. AmLaw 200 firms typically land at $280 to $320 per seat per month for Edge after volume discount — verify pricing at https://legal.thomsonreuters.com/en/products/westlaw/pricing as of June 2026.

**Lexis+ Essential** at $150 per seat per month gets you Shepards but no Protégé AI. **Lexis+ Advanced** at $200 adds Protégé. **Lexis+ AI Premium** at $250 unlocks the higher Protégé query limits and the AI drafting assistant. Lexis has been the aggressive price competitor since 2024 — they will match Westlaw on like-for-like seats and frequently throw in a six-month Practical Guidance trial. Mid-firms (50-200 lawyers) are paying $130 to $170 per seat per month for Advanced after negotiation.

**Bloomberg Law** at $450 per seat per month is the only flat-priced incumbent. Bloomberg does not tier its product, which is annoying for solos who want only BCite, but predictable for firms that want everything. The flat fee includes BCite, dockets, Practical Guidance, and the news service. Volume discounts kick in at 25 seats and Bloomberg will go to roughly $320 per seat per month at AmLaw scale — verify at https://pro.bloomberglaw.com/pricing/ as of June 2026.

**vLex Vincent AI** at $95 (Solo), $145 (Pro), and $195 (Enterprise) per seat per month is the genuine price disruptor (https://vlex.com/vincent-ai). The Solo tier is monthly with no commitment, which is unique in this market. The Pro tier adds API access and SSO. Enterprise adds custom data residency, audit logs, and dedicated support. Vincent is the only credible option for a mid-firm to test on a single practice group without a 12-month contract — that flexibility alone is worth the consideration.

**Casetext CARA AI inside CoCounsel** is the murky one. Thomson Reuters does not publish CoCounsel pricing and quotes vary from $400 to $750 per seat per month depending on which sub-features are included (https://casetext.com/cocounsel/). For firms that already pay for Westlaw Edge, CARA's brief-analysis features are partially available as Quick Check at no extra cost — which means the standalone CoCounsel pitch only makes sense if you want the full agentic workflow including deposition prep and document review.


Integration and workflow: where the citation checker meets your draft

The citation checker that lives inside your drafting environment is the one your associates actually use. The five vendors take meaningfully different approaches to Word, Outlook, and the browser, and the workflow friction matters more than any single feature comparison.

**Westlaw KeyCite** has the deepest Microsoft Word integration via Drafting Assistant, which is now standard in Westlaw Edge at no extra charge. Highlight a citation in a Word document, and Drafting Assistant returns the KeyCite flag, suggested replacement citations, and a one-click Bluebook reformat. The Quick Check button in the same toolbar runs CARA-style brief analysis. This is the most polished workflow on the market and the reason Westlaw retention rates are above 90 percent at AmLaw firms.

**Lexis Shepards** offers Lexis Create as the equivalent Word add-in, which launched in its current form in 2023 and reached parity with Drafting Assistant in late 2025. Create now embeds Protégé AI summaries directly in the Word margin, which is genuinely better than Westlaw's modal popup pattern. The Outlook integration is weaker than Westlaw's but improving. If your associates draft in Word and your partners read briefs in Outlook, Lexis Create is now the more productive surface.

**Bloomberg BCite** integrates through Bloomberg Drafting, a Word add-in that is competent but visibly newer than its competitors. The strength is the agency-citation handling — BCite is the only tool that will automatically link SEC, IRS, and FERC citations to the underlying regulatory text in Word. For transactional and regulatory drafts this is a productivity multiplier; for litigation it is overkill.

**vLex Vincent AI** integrates differently because Vincent is an agent, not a passive checker. The workflow is: write a question in plain English in the Vincent UI, Vincent drafts a memo with citations, then you copy the memo to Word. The Word add-in exists but is shallower than Westlaw's or Lexis's. For research-first workflows Vincent is faster than the incumbents; for citation-checking-only workflows it is slower because you are paying for the agent overhead you do not need.

**Casetext CARA AI inside CoCounsel** integrates via the CoCounsel Word add-in, which is the most aggressively AI-forward of the bunch. Upload a draft, CARA marks every citation, suggests additional cases, and proposes language edits — all in a single pass. This is the workflow brief-writers love and other lawyers find intrusive. The trade-off is that CoCounsel is a heavier install and a slower app than the incumbents (https://casetext.com/cocounsel/).


Decision matrix: which checker maps to which practice area

Practice area drives the right answer more than firm size does. The below mappings are what we recommend to partners who ask us directly which seat to buy, based on actual usage data from clients in 2025-2026. None of this is sponsored and we have no reseller relationship with any vendor.

Federal litigation in AmLaw 200 firms should buy **Westlaw Edge** at $350 to $500 per seat per month (https://legal.thomsonreuters.com/en/products/westlaw/pricing). The KeyCite federal coverage, the Drafting Assistant workflow, and the partner-comfort factor of the Westlaw brand at sanctions hearings are worth the premium. Add Bloomberg Law as a secondary seat for the dockets and news. This is the reference stack and most AmLaw 200 firms are running exactly this.

State court litigation in mid-size firms should buy **Lexis+ Advanced** at $200 per seat per month (https://www.lexisnexis.com/en-us/products/lexis-plus.page). Shepards state-court coverage genuinely beats KeyCite in California, Texas, New York, and Florida, and the Protégé AI summary at this price point is the best value in the market. The $150 per seat saved versus Westlaw funds a Bloomberg seat for the partner who needs regulatory work.

Transactional, M&A, capital markets, and securities defense should buy **Bloomberg Law** at $450 per seat per month flat (https://pro.bloomberglaw.com/pricing/). BCite is the only citator that treats agency materials as first-class citations, and the integration with Bloomberg's deal data, SEC filings, and news service is the differentiated workflow. KeyCite or Shepards alone will not catch the SEC interpretive letter that distinguishes your no-action argument.

International disputes, EU regulatory, and any cross-border practice should buy **vLex Vincent AI** at $145 to $195 per seat per month (https://vlex.com/vincent-ai). Vincent is the only tool with credible coverage across EU, UK, LATAM, and US jurisdictions in a single search, and the AI agent workflow makes multi-jurisdiction research dramatically faster. Pair Vincent with a single shared Westlaw or Lexis seat for US-deep work.

Brief-writing-heavy practices — appellate work, complex motion practice, sanctions defense — should add **Casetext CARA AI inside CoCounsel** at roughly $500 per seat per month (https://casetext.com/cocounsel/) on top of their primary research seat. CARA's brief-upload workflow finds missing citations and predicts opposing arguments at a level no incumbent matches. This is luxury and worth it only if briefs are your unit of production.


Security, data residency, and the AI training question every GC asks

Every general counsel evaluating legal AI tools in 2026 asks the same three questions: where does my privileged data go, who trains on it, and what is the audit trail. The answers vary meaningfully across these five vendors and the differences are now an explicit factor in procurement scoring at AmLaw 100 firms.

**Westlaw** stores customer queries and uploaded documents in US, EU, or UK regions depending on contract, with the Edge tier defaulting to US East. Thomson Reuters published in late 2024 that customer queries are not used to train Westlaw's foundation models without explicit opt-in, which most firms decline. The audit log is exposed via the Westlaw admin console and via a SCIM-based API at the Edge enterprise tier (https://legal.thomsonreuters.com/en/products/westlaw/pricing).

**Lexis Shepards and Lexis+** offer US, EU, UK, and Australia regions, which is the broadest residency footprint of the incumbents. Lexis was more aggressive than Westlaw in publishing a no-training-on-customer-data commitment, which it did at Lexis+ AI launch in 2024. The Protégé AI layer runs on Anthropic Claude via a private deployment, which means OpenAI is not in the data path — a meaningful detail for European clients.

**Bloomberg Law** is US-only on data residency, which is the limitation that excludes Bloomberg from most European mandates. BCite AI Insights runs on Bloomberg's in-house model, which is differentiated from the foundation-model dependence of competitors. For US firms with US-only data this is a strength; for any multinational practice it is a hard limit (https://pro.bloomberglaw.com/pricing/).

**vLex** is EU-primary on residency, which is unique among credible legal AI vendors. Vincent's data path runs through EU AWS regions by default with US East available on Pro and Enterprise tiers. vLex has been more transparent than the incumbents about which third-party foundation models it uses (a combination of OpenAI, Anthropic, and Mistral) and publishes per-feature model assignments. For European firms this is the natural choice.

**Casetext CARA AI inside CoCounsel** inherits Thomson Reuters's US-East default residency, which is a step back from standalone Casetext circa 2022 when Casetext offered more flexible options. CoCounsel's audit logging is the most granular on the market — every query, every retrieved case, every model invocation is logged and exposed via API — which has made CoCounsel popular at financial services in-house teams.


What an honest one-year total cost looks like for each tool

Sticker price per seat per month is misleading because it ignores annual minimums, integration costs, and the inevitable add-ons. Below is what a 10-attorney litigation boutique should expect to pay in year one for each option, based on confirmed invoices from clients we work with as of June 2026.

**Westlaw Edge at AmLaw-comparable specs** for 10 seats lands at roughly $42,000 to $52,000 per year after the typical 30 percent volume discount on the $350 to $500 list. Add $0 for Drafting Assistant (included), $0 for Quick Check (included on Edge), and roughly $8,000 for a Practical Law subscription if you want secondary materials. All-in: $50,000 to $60,000 per year. This is the budget number partners should plan around.

**Lexis+ Advanced for 10 seats** lands at $20,000 to $24,000 per year after volume discount on the $200 list. Add $0 for Lexis Create (included), $0 for Protégé Standard (included on Advanced), and $6,000 to $9,000 for Practical Guidance. All-in: $26,000 to $33,000 per year — roughly half the Westlaw spend for comparable coverage on most state-litigation work (https://www.lexisnexis.com/en-us/products/lexis-plus.page).

**Bloomberg Law for 10 seats** lands at $48,000 to $54,000 per year — Bloomberg is the least negotiable on price below 25 seats. The trade-off is the flat-fee predictability and the included dockets, deal data, news service, and Practical Guidance. All-in: $48,000 to $54,000. Bloomberg is the right number for transactional firms that would otherwise pay separately for dockets and news.

**vLex Vincent AI Pro for 10 seats** lands at $14,000 to $18,000 per year on the $145 list with the typical 15 percent annual-prepay discount. No annual minimum is required — this is the only vendor on this list that allows true month-to-month at scale. All-in: $14,000 to $18,000 (https://vlex.com/vincent-ai). For a litigation boutique that does not need the AmLaw-brand factor, Vincent is one-quarter the Westlaw cost.

**Casetext CARA AI inside CoCounsel for 10 seats** lands at $48,000 to $72,000 per year on the $500 list, depending on which CoCounsel modules you include. This is on top of, not in place of, your primary research seat — which is why CoCounsel is best understood as a brief-writing premium add-on rather than a primary citation checker. All-in incremental: $48,000 to $72,000 (https://casetext.com/cocounsel/).


The 9/10 honest take: which to buy if we were starting a firm tomorrow

If we were standing up a litigation boutique in June 2026 and we had to pick one citation checker for the whole firm, we would buy **Lexis+ Advanced** at $200 per seat per month. The Shepards dataset is competitive with KeyCite on federal work and better on state work, the Protégé AI summary is the best price-performance AI layer in the market, and the $150 per seat saved versus Westlaw Edge funds either a Vincent seat for international work or a Bloomberg seat for the regulatory partner. This is the honest answer and most partners will not like it because they grew up on Westlaw (https://www.lexisnexis.com/en-us/products/lexis-plus.page).

If the firm is AmLaw 100 and partner comfort and judicial-clerk training matter — buy **Westlaw Edge** anyway. The Drafting Assistant workflow, the Quick Check brief analysis included on Edge tiers, and the simple fact that most federal judges grew up reading KeyCite flags make Westlaw the safe choice for high-stakes federal litigation. Pay the premium and stop second-guessing it (https://legal.thomsonreuters.com/en/products/westlaw/pricing).

If the practice is transactional or regulatory — buy **Bloomberg Law** as the primary seat. BCite is the only citator on this list that treats agency materials as first-class. Yes, $450 per seat per month is the highest sticker price in the group, but the bundled dockets, deal data, news, and Practical Guidance mean you are paying for one tool rather than three (https://pro.bloomberglaw.com/pricing/).

If the firm is international, EU-anchored, or cost-sensitive at startup — buy **vLex Vincent AI**. At $95 to $195 per seat per month with no annual minimum, Vincent is the only tool that lets a 3-attorney firm get AmLaw-grade research capability without a $50,000 annual commitment. The US-treatment-graph thinness is a real limitation; the price-performance is unmatched everywhere else (https://vlex.com/vincent-ai).

If brief-writing is the unit of production — buy **Casetext CARA AI inside CoCounsel** as an add-on, not a primary tool. Pair it with Lexis+ Advanced as your primary research seat. Total cost lands at roughly $700 per seat per month combined, which is still less than Westlaw Edge plus Quick Check at most firms, and you get the best brief-analysis workflow on the market (https://casetext.com/cocounsel/).

The meta-point: there is no single right answer because there is no single research workflow. Run a 14-day trial on at least two vendors before signing a 12-month contract. Three of the five vendors above (Lexis, Bloomberg, vLex) offer full-feature trials. Westlaw and CoCounsel offer demo-only — push your account rep for a real trial and walk away if they refuse.

How to pick between Westlaw KeyCite, Lexis Shepards, Bloomberg BCite, vLex Vincent, Casetext CARA AI for your team

  1. 1

    Map your practice mix to the right primary dataset

    Before you talk to any account rep, write down the percentage split of your firm's work across federal litigation, state litigation, transactional, regulatory, and international. If federal litigation is over 60 percent of revenue, Westlaw or Lexis. If transactional or regulatory is over 40 percent, Bloomberg moves to the top. If international is over 25 percent, vLex Vincent is the only credible option. This sounds obvious and most firms skip it — they buy what the senior partner used at her last firm and pay an extra $30,000 a year for the wrong tool. Spend an hour on the practice-mix exercise before you spend $50,000 on a contract.

  2. 2

    Demand a real 14-day trial on at least two vendors

    Lexis+ (https://www.lexisnexis.com/en-us/products/lexis-plus.page), Bloomberg Law (https://pro.bloomberglaw.com/pricing/), and vLex Vincent (https://vlex.com/vincent-ai) all offer full-feature trials. Westlaw and Thomson Reuters CoCounsel offer demo-only — push back and demand a 14-day full-feature trial, and if they refuse, that is itself information. Pick three real briefs from the last six months and run them through each tool. Track time-to-first-meaningful-result, false-negative count on known-bad cases, and which AI summary your senior associates trust without re-reading the case. The right answer is rarely the prettiest UI.

  3. 3

    Negotiate annual seats, not list price

    Every vendor publishes a website price that is the floor for solos and the starting point for negotiation at firms with five or more seats. Realistic discount targets in 2026: Westlaw 25-40 percent off list for 10+ seats, Lexis 25-35 percent off list for 10+ seats, Bloomberg 15-30 percent off list at 25+ seats. vLex is the only vendor that does not aggressively discount because their list price is already the disruptor price. Get quotes from at least two incumbents in writing before signing — the second quote is what moves the first.

  4. 4

    Lock the data-residency and no-training clauses in writing

    The vendor sales deck will tell you customer data is not used for foundation-model training. The contract may say something different in the small print. Insist on a contractual no-training commitment for the AI features specifically — Lexis and vLex publish standard clauses; Westlaw and Bloomberg will provide one on request. If you have European clients, lock data residency in the contract too. The June 2026 enforcement reality is that GCs are auditing this annually and pulling work from firms that cannot prove their stack is compliant.

  5. 5

    Re-evaluate every twelve months and stop being loyal

    Legal research pricing and AI features moved more between January 2024 and June 2026 than they did in the prior decade. Lexis Protégé did not exist in early 2024. Westlaw Edge AI did not exist in early 2024. vLex Vincent was beta. CoCounsel was a standalone product. Set a calendar reminder for month ten of every 12-month contract to re-bid the seat. The 2026 winners are not necessarily the 2027 winners, and the incumbents only sharpen pricing when they think you might leave. Loyalty to a research provider costs roughly $1,000 to $2,000 per seat per year and buys nothing your associates care about.

Frequently Asked Questions

Is Westlaw KeyCite still more accurate than Lexis Shepards in 2026?

Not uniformly. KeyCite is still ahead on federal appellate negative-treatment recency by roughly one to two weeks on average, which matters for high-stakes brief filing. Shepards is ahead on state appellate negative treatment in California, Texas, New York, and Florida, where the Lexis editorial team has historically been deeper. For AI summaries of why a case was distinguished, Lexis Protégé (https://www.lexisnexis.com/en-us/products/lexis-plus.page) is competitive with Westlaw's AI-Assisted Research and meaningfully cheaper at $200 per seat per month versus $350+ for Westlaw Edge. The honest answer is that both miss things and high-stakes briefs deserve cross-checking on at least two citators.

What does Casetext CARA AI cost as a standalone product in 2026?

It does not exist as a standalone product anymore. Thomson Reuters acquired Casetext in 2023 and folded CARA into the CoCounsel platform, which is priced at roughly $500 per seat per month for the Core tier and up to $650+ for Premium (https://casetext.com/cocounsel/). Firms that already subscribe to Westlaw Edge get a slimmed-down version of CARA's brief-analysis workflow as Quick Check at no additional cost. If you want the full CARA suggestion engine and document review features, you are buying CoCounsel — verify pricing at casetext.com/cocounsel as of June 2026 because Thomson Reuters does not publish list pricing publicly.

Is vLex Vincent AI good enough to replace Westlaw or Lexis for a US-only practice?

For solo and small-firm US-only litigation, yes — with one caveat. Vincent's US treatment-graph coverage is genuinely competitive in 2026 for federal cases and major state jurisdictions, and at $95 to $195 per seat per month (https://vlex.com/vincent-ai) it is one-third the cost of comparable Westlaw or Lexis seats. The caveat is partner comfort and judicial credibility — federal judges and clerks expect to see Westlaw KeyCite citations in motion practice, and switching to Vincent introduces a small risk that opposing counsel or the court questions an unfamiliar citator. For AmLaw 200 work the safe choice is still Westlaw or Lexis; for everything else Vincent is genuinely competitive.

Why is Bloomberg Law $450 per seat per month when Lexis starts at $150?

Bloomberg Law is a flat-priced bundle that includes BCite plus dockets, deal data, the news service, Practical Guidance, and the broader Bloomberg Terminal-adjacent data. Lexis+ Essential at $150 per seat per month gets you Shepards but not the equivalent transactional, news, or regulatory bundles (https://www.lexisnexis.com/en-us/products/lexis-plus.page versus https://pro.bloomberglaw.com/pricing/ as of June 2026 — verify at each vendor's pricing page). For a litigation-only practice the Lexis price is the honest comparison; for a transactional practice the Bloomberg bundle replaces three to four separate subscriptions and the all-in cost is competitive.

Do any of these tools train AI models on my privileged client data?

All five vendors have published commitments that customer data and queries are not used to train shared foundation models without explicit opt-in. Lexis (https://www.lexisnexis.com/en-us/products/lexis-plus.page) and vLex (https://vlex.com/vincent-ai) were earliest and most aggressive in publishing standard contract language. Westlaw and Bloomberg will provide a no-training clause on request. Thomson Reuters CoCounsel includes the commitment in its standard MSA. Get the commitment in writing in the signed contract, not in the marketing FAQ — the legal teams at all five vendors will provide it, and any reluctance is itself a procurement red flag.

Can I use just one citation checker or do I need a backup?

For low-stakes work — research memos, internal advice, client updates — one citator is sufficient and adding a second adds cost without meaningful risk reduction. For high-stakes work — federal appellate briefs, sanctions-risk motions, dispositive motions in complex litigation — most AmLaw 100 firms run citations through two independent citators because the false-negative rate on any single one is non-zero. The cheapest credible two-citator stack in 2026 is Lexis+ Advanced at $200 per seat plus a shared Westlaw Precision seat at $200 per month for the brief filer — roughly $250 per attorney per month all-in for a 10-attorney firm.

How long does a real Lexis or Westlaw negotiation take and when should I start?

Start 90 days before your existing contract renews. Both Lexis and Westlaw run a 60-day negotiation cycle at firms with 10 to 50 seats and longer at AmLaw scale. Get a competing quote from the other incumbent in writing in week one — this is the single highest-leverage action and it routinely produces 20 to 30 percent off list versus negotiating without a comparison quote. Add a vLex Vincent quote (https://vlex.com/vincent-ai) as a third reference point even if you are not seriously considering switching — the existence of credible $150-per-seat competition meaningfully changes incumbent flexibility on price.

What is the cheapest credible citation-checker stack for a solo practitioner in 2026?

vLex Vincent Solo at $95 per seat per month (https://vlex.com/vincent-ai) is the cheapest credible option for a US-only litigation solo and it includes citation checking, AI research, and Word integration. For a solo who does federal court work and wants the partner-comfort factor, Lexis+ Essential at $150 per seat per month (https://www.lexisnexis.com/en-us/products/lexis-plus.page) is the next step up and includes Shepards but not the Protégé AI summaries. Westlaw Precision at $200 per seat per month is the equivalent Westlaw entry point — verify at vendor.com/pricing as of June 2026. Avoid Bloomberg and CoCounsel as solo options; both are priced for firms with five-plus seats.

Does any tool catch citations that the others miss in 2026?

Yes, regularly, and this is the actual case for a two-citator workflow on high-stakes work. KeyCite catches federal appellate negative treatment fastest. Shepards catches state appellate criticism in specialty reporters that Westlaw indexes later. BCite catches agency and regulatory cross-citations the case-law-first citators miss entirely. Vincent catches international and cross-jurisdictional citations the US-only tools never see. CARA's brief-analysis engine catches cases you should have cited but did not — a different kind of catch entirely. No single tool is comprehensive and any vendor that tells you otherwise is selling, not serving.

You now know which citation checker to buy. Now make every prompt those tools run hit harder.

AI Prompt Generator builds production-ready system prompts that work across ChatGPT, Claude, Gemini, and every tool in this article — including Lexis Protégé, Westlaw AI-Assisted Research, Bloomberg BCite Insights, vLex Vincent, and Casetext CARA. Stop wasting $500-per-seat AI features on lazy one-line queries and start shipping prompts that produce defensible work product on the first try. 14-day free trial, no credit card required.

Browse all prompt tools →