TL;DR: CoCounsel is the original AI legal assistant. It launched in 2023, was bought by Thomson Reuters for $650M four months later, and crossed 1 million users across 107 countries by February 2026, according to TR. Its real moat is the one thing no model can replicate: native, citation-grounded access to Westlaw and Practical Law.
On the independent 50-point benchmark we publish, CoCounsel scored 37/50 on the generic legal evaluation, against 49/50 for HAQQ. Across all 11 task categories it averaged 36.2 to HAQQ's 47.5, and never finished above fifth.
The benchmark scores standalone answers, which is exactly where CoCounsel's Westlaw moat does not show up. So this review credits the integration and lets the numbers argue everything else. If you live inside Westlaw, CoCounsel is the obvious add-on. If you do not, or you work in Arabic or civil-law MENA, the value math changes fast. Data below.
Most CoCounsel reviews are written by people who have never run a prompt through it, or by competitors grading their own competitor. This one is different in one specific way: it is backed by published benchmark scores. We test CoCounsel and 18 other models and platforms across 11 legal task categories on a public 50-point rubric, and the full table is on our compare page for anyone to check.
Disclosure first, because it matters: 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, cited as such, or from a named external source you can click.
Key facts about CoCounsel
- Launched March 1, 2023 by Casetext as "the first AI legal assistant," powered by exclusive early access to OpenAI's GPT-4, according to Casetext's launch announcement (PR Newswire, 2023).
- Acquired by Thomson Reuters for $650 million in cash; the deal signed June 26, 2023 and closed August 17, 2023, according to Thomson Reuters (2023).
- Crossed 1 million users across 107 countries and territories by February 24, 2026, three years after launch, according to LawSites / Thomson Reuters (2026).
- Now runs a multi-model architecture using frontier models from Anthropic, OpenAI and Google plus TR's own AI, integrated across Westlaw, Practical Law, Checkpoint and Microsoft 365, according to LawSites (2026).
- CoCounsel Legal, the agentic relaunch, shipped August 5, 2025 with Deep Research that plans, retrieves from Westlaw and Practical Law in parallel, and verifies citations, according to Thomson Reuters (2025).
- Pricing is sold as a Westlaw add-on, not a clean price sheet: third-party 2026 analyses put CoCounsel Essentials plans roughly $104 to $639 per user per month, before counting the underlying Westlaw subscription.
- 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: 36.2 vs 47.5.
- Weakest benchmark category: law explanation, at 31/50, where plain ChatGPT outscored it by 11 points.
What is CoCounsel?
CoCounsel is an AI legal assistant for research, document review, drafting, deposition prep, and contract analysis. Casetext launched it on March 1, 2023 as the first product of its kind, built on early, exclusive access to GPT-4 before that model was public, according to Casetext's announcement (PR Newswire, 2023).
Thomson Reuters bought Casetext for $650 million in cash roughly four months later, signing on June 26 and closing on August 17, 2023, according to Thomson Reuters (2023). That acquisition is the whole story of CoCounsel today. It stopped being a standalone startup tool and became the AI front end for Westlaw, TR's flagship legal research database.
By scale, CoCounsel is a category leader. TR says it passed 1 million users across 107 countries by February 2026, three years after launch, according to LawSites (2026). A serious review starts there, because a review that pretends the incumbent's product is bad at everything is not a review, it is an ad.
Westlaw and Casetext: the lineage that defines CoCounsel
You cannot understand CoCounsel without understanding what owns it. Thomson Reuters owns Westlaw and Practical Law, two of the most authoritative legal content sets in the common-law world. Westlaw's editorial layer includes the West Key Number System, a taxonomy that has organized US case law for over a century. That content is the asset. CoCounsel is the interface to it.
In August 2025 TR relaunched the product as CoCounsel Legal, an agentic system with a Deep Research mode that, per TR, builds multi-step research plans, retrieves from Westlaw and Practical Law in parallel, traces its reasoning, and returns citation-backed reports, according to Thomson Reuters (2025). The TR Institute describes it as re-architected around Anthropic's Claude Agent SDK at the model layer, giving it native access to Westlaw and Practical Law with a citation-ledger architecture that tracks every source the agent reads, according to the Thomson Reuters Institute (2025). By February 2026 TR described CoCounsel as running a multi-model architecture across Anthropic, OpenAI and Google models, according to LawSites (2026).
This is the part that does not show up in a standalone-answer benchmark, and we will say it plainly: grounding generation in Westlaw and Practical Law, with a citation ledger built into the system rather than bolted on, is a genuine answer to the legal AI hallucination problem. For a US litigator who already pays for Westlaw, that is the single best reason to add CoCounsel.
CoCounsel benchmark scores: 37/50 vs the field
The independent benchmark we publish on our compare page scores 19 models and platforms, from frontier models like Claude and ChatGPT to legal verticals like Harvey, CoCounsel, and Spellbook. The generic evaluation is a 50-point rubric covering Sharia, statute, forum, clause, risk, hallucination, formatting, brevity, partner-readiness, and source linking. Ten further categories score specific deliverables, from NDAs to shareholder agreements.
Here is CoCounsel against HAQQ across all 11 categories:
| Task category | CoCounsel /50 | HAQQ /50 | Gap |
|---|---|---|---|
| Generic legal evaluation | 37 | 49 | 12 |
| Contract drafting | 36 | 47 | 11 |
| Legal research | 36 | 48 | 12 |
| Law explanation | 31 | 46 | 15 |
| Employment agreement | 37 | 48 | 11 |
| Professional memorandum | 39 | 46 | 7 |
| License agreement | 35 | 47 | 12 |
| Shareholder agreement | 37 | 48 | 11 |
| Consultancy agreement | 36 | 47 | 11 |
| Commercial agreement | 37 | 48 | 11 |
| NDA drafting | 37 | 49 | 12 |
| Average | 36.2 | 47.5 | 11.3 |
What the table actually says:
- CoCounsel's best category is professional memorandum, at 39/50, its only top-five finish. Memos are research-and-synthesis work, the home turf of a Westlaw-grounded tool, and the score reflects it.
- Its weakest category is law explanation, at 31/50. Plain ChatGPT scored 42 on the same task. If your job is explaining the law in plain language to a non-lawyer client, a free chatbot beat the Thomson Reuters platform by 11 points on our rubric.
- On legal research, CoCounsel's 36 trails its own sibling, LexisNexis +AI (41), and Perplexity Sonar (38). Our rubric scores the standalone answer, not the live Westlaw retrieval behind a real CoCounsel session, which is the single biggest caveat on this table. See the honesty note below.
- Raw frontier models outscore it on answer quality. Claude Fable 5 scored 45 on the generic evaluation and Claude Opus 4.7 scored 42, against CoCounsel's 37, and the ordering repeats across categories. What you pay CoCounsel for is not the raw answer. It is Westlaw underneath the answer.
- Within its peer group it is mid-pack, not the leader. Harvey edged it on the generic evaluation, 38 to 37, and Harvey leads it on most drafting categories. The vertical-platform crown in our test goes to Harvey, then CoCounsel and Lexis tied a notch below.
Honesty check, twice over. First: we publish this benchmark, and HAQQ ranks first in it. Treat that the way you treat any vendor-published number with public methodology. The rubric and all 19 scores are on our compare page. Second, and specific to CoCounsel: our rubric grades a standalone answer, so it does not exercise CoCounsel's live Westlaw and Practical Law retrieval, its citation ledger, or its agentic Deep Research workflow. Those are real CoCounsel strengths this table cannot measure. Read the research-category number as a floor, not a verdict.
CoCounsel pricing: what it really costs
Thomson Reuters does not publish a clean CoCounsel price sheet. It is sold as an add-on to Westlaw, and the number you pay depends on firm size, your existing Westlaw contract, and which content tiers you carry.
Third-party 2026 pricing analyses put the CoCounsel Essentials and Westlaw Advantage plans in a wide band, roughly $104 to $639 per user per month, with the high end reflecting solo attorneys carrying all-states-and-federal Westlaw coverage on annual terms. The structural catch is that those figures sit on top of the underlying Westlaw subscription, so several of those analyses peg a realistic all-in cost at $300 to $600+ per user per month once Westlaw itself is counted.
None of that is a knock on the model. Bundling an AI assistant with the research database it depends on is coherent, and the 1-million-user number says the market accepts it. But it means CoCounsel's price is effectively a Westlaw price. If you are not already a Westlaw firm, you are not buying an AI add-on, you are buying into the whole Thomson Reuters stack. For a solo in Dubai, a three-lawyer civil-law firm in Beirut, or a founder who needs one shareholder agreement reviewed, that is a different and much larger decision.
CoCounsel's price is really a Westlaw price. The AI is the easy part of that bill to swallow. The database underneath it is the rest.
When CoCounsel is the right choice
Fair is fair. CoCounsel is a defensible, often correct choice if most of these are true:
- You are a US or UK firm whose research already runs on Westlaw and Practical Law.
- Your work is primarily common-law, English-language research and litigation.
- Citation grounding matters more to you than raw answer fluency, and you want retrieval traceable to authoritative sources.
- You value the deployment maturity of a Thomson Reuters product over unit economics.
- You want one agentic tool that plans across research, drafting, and document review without you stitching steps together.
If that is you, CoCounsel belongs on your shortlist, and the Westlaw integration is a real reason to keep it there over a generic chatbot. The benchmark gap above is about standalone answer quality, which is not the dimension you would be buying CoCounsel for.
Where CoCounsel falls short
Lock-in: the Westlaw moat is also a Westlaw cage
The integration that makes CoCounsel strong is the same thing that traps you. Its best work is grounded in Westlaw and Practical Law, content you license from Thomson Reuters and lose access to the day you leave. You are not adopting a portable AI assistant. You are deepening a dependency on a single vendor's content stack, at that vendor's pricing power. That is great until renewal season.
Language and jurisdiction: common-law-first, English-first
Westlaw and Practical Law are built for US, UK, and adjacent common-law systems. CoCounsel inherits that footprint. For Arabic-language work, or civil-law MENA jurisdictions like the UAE mainland, Saudi, Egypt, or Lebanon, the authoritative content layer that makes CoCounsel valuable in the US is thin or absent. We mapped this gap in our MENA legal tech market coverage.
Standalone answer quality trails frontier models
On our benchmark, raw Claude outscored CoCounsel on all 11 categories, and ChatGPT beat it on law explanation. CoCounsel's value is the Westlaw grounding, not the underlying reasoning, which it increasingly rents from the same frontier labs anyone can call. Strip the database away and the model layer is not a differentiator.
CoCounsel alternatives by use case
| Use case | Strongest fit | Why |
|---|---|---|
| MENA, Arabic, civil-law jurisdictions | HAQQ Legal AI | Native Arabic with RTL, 80+ countries, 49/50 generic benchmark |
| Solo lawyers, SMBs, consumers | HAQQ Legal AI | Self-serve, free tier, $33 to $100/mo, no database lock-in |
| BigLaw English common-law platform | Harvey | Largest deployed footprint, 38/50 generic, enterprise workflows |
| Collaborative review at European firms | Legora | $100M+ ARR, $5.6B valuation, Arabic UI via Al Tamimi |
| Contract work inside Microsoft Word | Spellbook | Deep Word integration, focused transactional tooling |
| US Lexis-stack research | Lexis+ AI | Lexis content moat; scored 41/50 on legal research, beat CoCounsel |
| Build-your-own on a frontier model | Claude | 45/50 generic at API prices, but you own verification |
HAQQ Legal AI: for MENA, Arabic, and everyone outside the Westlaw stack
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 benchmark and ranked first in all 11 categories, and unlike most of the market we disclose our engine, Justinian. Where CoCounsel's value depends on a US-and-UK content stack you license forever, HAQQ is built for the Arabic-language, civil-law work that stack barely touches. You can test it free in minutes, no Westlaw contract required.
Prueba HAQQ AI gratis
Experimenta la redacción e investigación legal con IA
Harvey: the BigLaw platform CoCounsel competes with
Harvey is the enterprise legal AI leader by traction and edged CoCounsel on our generic evaluation, 38 to 37. It is English-first, common-law-first, and enterprise-only, with no public pricing. If you are a large firm choosing between the two, the decision is largely whether you want to deepen a Thomson Reuters relationship or build on Harvey's workflow platform. We broke Harvey down in our Harvey review.
Lexis+ AI: the other research incumbent
CoCounsel's closest mirror is LexisNexis +AI, the same play built on the Lexis content stack instead of Westlaw. On our benchmark Lexis+ AI actually beat CoCounsel on legal research, 41 to 36, which says the choice between them is mostly a choice between which research database you already pay for. If you are a Lexis firm, this is the natural pick.
Spellbook: if your whole life is in Word
Spellbook scored 25/50 generic and 34/50 on contract drafting, modest numbers, but the benchmark scores standalone answers and Spellbook's real value is living inside Microsoft Word where transactional lawyers already work. For a focused contracts team that does not need Westlaw research, it is a cheaper and more focused tool.
Raw frontier models: the DIY route
Claude Fable 5 scored 45/50 on the generic evaluation, eight points above CoCounsel, at API prices. The catch is the verification burden CoCounsel's citation ledger is designed to solve: 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. Go this route only if you are prepared to build the citation-verification and jurisdiction-governance layer yourself, which is precisely the layer you would otherwise be paying Westlaw for.
Key takeaways
- CoCounsel is the original AI legal assistant, now a Thomson Reuters product: launched 2023, acquired for $650M, 1M+ users across 107 countries by Feb 2026, according to TR and LawSites (2026).
- Its real moat is Westlaw and Practical Law grounding, with an agentic Deep Research mode and citation ledger, the part a standalone-answer benchmark cannot measure.
- On published benchmark scores it is mid-pack among legal verticals: 37/50 generic vs HAQQ's 49/50, never above fifth, weakest on law explanation at 31.
- Pricing is really Westlaw pricing: sold as an add-on, roughly $104 to $639 per user per month per third-party 2026 estimates, on top of the underlying database subscription.
- Pick by use case: CoCounsel for Westlaw-stack US common-law research; HAQQ for MENA, Arabic, and the consumer and SMB market outside the database moat; Harvey, Lexis+ AI, and Spellbook for their respective lanes.
FAQ
What is CoCounsel?
CoCounsel is an AI legal assistant for research, drafting, document review, deposition prep, and contract analysis. Casetext launched it in March 2023 as the first AI legal assistant, built on early access to GPT-4, and Thomson Reuters acquired Casetext for $650 million the same year, according to Casetext and Thomson Reuters (2023). It is now integrated across Westlaw, Practical Law, and Microsoft 365.
How much does CoCounsel cost?
Thomson Reuters publishes no clean price sheet. CoCounsel is sold as a Westlaw add-on, and third-party 2026 analyses put the Essentials and Westlaw Advantage plans in a roughly $104 to $639 per user per month band, with the high end reflecting full Westlaw coverage on annual terms. Counting the underlying Westlaw subscription, several analyses estimate a realistic all-in of $300 to $600+ per user per month.
How did CoCounsel score on an independent benchmark?
On the 50-point benchmark published on our compare page, CoCounsel scored 37/50 on the generic legal evaluation versus 49/50 for HAQQ, and averaged 36.2 across all 11 task categories versus HAQQ's 47.5. Its best category was professional memorandum (39) and its weakest was law explanation (31). The benchmark scores standalone answers, so it does not exercise CoCounsel's live Westlaw retrieval or citation ledger.
Is CoCounsel the same as Casetext?
CoCounsel was built by Casetext, the company. Thomson Reuters acquired Casetext for $650 million in 2023, so CoCounsel is now a Thomson Reuters product integrated with Westlaw and Practical Law. You will sometimes see it referred to as Casetext CoCounsel for that reason.
Does CoCounsel work with Westlaw?
Yes, and that is its core value. CoCounsel Legal, relaunched in August 2025, has native access to Westlaw and Practical Law and grounds its research and drafting in that content with a citation-backed Deep Research workflow, according to Thomson Reuters (2025). Its pricing is structured as a Westlaw add-on, so in practice you need the Westlaw stack to get the full benefit.
What are the best CoCounsel alternatives?
It depends on the use case: HAQQ for MENA, Arabic, civil-law jurisdictions, and any solo, SMB, or consumer user; Harvey for BigLaw English common-law platforms; Lexis+ AI if you are on the Lexis research stack instead of Westlaw; Spellbook for Word-native contract work; and a raw frontier model like Claude if you can build your own verification layer.
Does CoCounsel support Arabic or MENA law?
Not meaningfully. CoCounsel's value comes from Westlaw and Practical Law, which are built for US, UK, and adjacent common-law systems. For Arabic-language work or civil-law MENA jurisdictions like the UAE mainland, Saudi, Egypt, or Lebanon, that authoritative content layer is thin or absent. HAQQ ships native Arabic with RTL across 80+ countries today.
Is CoCounsel worth it?
For a US or UK firm already on Westlaw, often yes: the citation-grounded research is a genuine answer to AI hallucination and a real reason to choose it over a generic chatbot. The premium buys the database and the grounding, not raw answer quality, where frontier models scored higher in our test. For firms outside the Westlaw stack, or working in Arabic and civil-law MENA, the lock-in and price decide the question before features do.
- HAQQ vs the field: full benchmark table
- Try HAQQ Legal AI free
- Best AI for legal work: 300-task frontier benchmark
- Harvey AI review and alternatives
- Best legal AI tools 2026
- Legora vs HAQQ: comparative analysis
- Thomson Reuters completes acquisition of Casetext ($650M, 2023)
- Casetext launches CoCounsel, the first AI legal assistant, on GPT-4 (PR Newswire, 2023)
- CoCounsel reaches 1 million users, three years after launch (LawSites, Feb 2026)
- Thomson Reuters launches CoCounsel Legal with agentic AI and Deep Research (Aug 2025)
- CoCounsel Legal reimagined (Thomson Reuters Institute, 2025)
- CoCounsel Legal plans and pricing (Thomson Reuters)
