✓ Reviewed by The Future Signal
✓ Reviewed by The Future Signal
Consensus answers research questions from over 200 million peer-reviewed papers, including all of PubMed, and structurally cannot fabricate a citation because it only summarises what it retrieves. Its Consensus Meter gives an immediate read on where evidence sits. It is the easiest research tool in its category and one of the cheapest, particularly with the 40% academic discount. Two caveats: pricing roughly doubled during 2026, and it deliberately does not support systematic reviews or document upload.
Consensus reduces the time between having a question and having a defensible answer. For clinicians, analysts and decision-makers, checking what published research actually says has traditionally meant database searches and abstract reading measured in hours; Consensus returns a cited synthesis in seconds.
The structural inability to fabricate sources removes the verification burden that makes general AI assistants risky for professional use. For businesses, the practical value is decision quality: it makes checking whether a claim in a vendor pitch, industry report or internal proposal is actually supported by evidence cheap enough that people will do it.
Compare plans and pricing to find the best option for your needs.
Compare plans and pricing to find the best option for your needs.
Last Updated: 28 August 2026
Consensus answers research questions from peer-reviewed papers rather than from training data, which means it structurally cannot invent a citation. Ask whether something works and it returns a Consensus Meter showing how many studies agree, with every claim linked to a real study. It is fast, genuinely easy to use, and covers more than 200 million papers including all of PubMed. Two things buyers should know: it deliberately does less than Elicit, and its price roughly doubled during 2026 — a change most published guides have not caught up with.
Estimated reading time: 9 minutes
Consensus is an AI search engine for scientific literature, built by Consensus NLP, Inc. You ask a research question in plain language and it returns an answer synthesised from actual peer-reviewed studies, with every claim linked back to the paper it came from.
The corpus runs to more than 200 million academic documents, aggregated from Semantic Scholar, OpenAlex and Consensus’s own scholarly crawl. Coverage includes nearly all high-impact journals and the entirety of PubMed, which makes it particularly strong for health and life sciences.
The architectural choice that defines it is what it will not do. Consensus does not answer from a model’s internal knowledge. It retrieves papers first, then summarises only what those papers say, with checker models verifying relevance before summarisation. A general assistant can invent a plausible-looking citation; Consensus cannot, because it has nothing to invent from.
Its signature output is the Consensus Meter. Ask a yes/no research question and it shows how the retrieved studies distribute across yes, no and possibly — a fast read on where evidence actually sits rather than what the loudest paper claims.
Not to be confused with: there is a separate company called Consensus that sells demo automation software, with prices in the hundreds of dollars per month. If you see figures like $600 or $1,250 a month attached to “Consensus pricing,” you are looking at the wrong product. This review covers consensus.app, the research tool.
Consensus is a strong fit for:
Consensus is probably not the right choice for:
Consensus Meter. The standout feature. For yes/no research questions, it displays how retrieved studies distribute across supporting, opposing and mixed findings — a quick, honest read on where evidence sits.
Pro messages. AI-generated summaries drawing on up to 20 papers from a search, unlimited on paid plans. This is the everyday workhorse of the product.
Deep reviews. A more thorough analysis across up to 50 papers, aimed at genuinely exploring a complex question. Allocated monthly rather than unlimited: 15 a month on Pro, 200 on the Deep plan.
Study Snapshots. One click on any paper shows key details — methodology, sample size, study duration — without opening the full text. Unlimited on paid plans.
Ask Paper. Interrogate an individual paper directly with questions about methods, findings or implications.
Advanced filters. Narrow by publication date, study design, journal quality and sample size, which is what separates this from a general search box.
Reference manager integration. Works with Zotero, Mendeley and EndNote, with LibKey support for reaching paywalled articles through institutional library access.
Multilingual queries. Ask in a language other than English and still search the corpus effectively.
QUESTION
│
▼
SEARCH ────────► ✓ 200M+ papers, all of PubMed
│ ✗ no news, web, industry reports
▼
EVIDENCE READ ─► ✓ Consensus Meter, Pro messages
│ ✓ Study Snapshots, filters
▼
DEEP DIVE ─────► ✓ Deep reviews, up to 50 papers
│ (metered: 15/mo Pro, 200/mo Deep)
▼
SYSTEMATIC ────► ✗ No PRISMA workflow, no screening
REVIEW pipeline, no structured extraction
│ → use Elicit
▼
YOUR DOCS ─────► ✗ No upload of your own files[Illustration placeholder: Consensus coverage map]
This is the easiest research tool in its category, and that is the point. There is no workflow to configure, no extraction columns to write, no prompt engineering. You type a question and read an answer with the studies underneath it.
That simplicity is a deliberate trade. Elicit asks more of you and gives more back in structure. Consensus asks almost nothing and gives you a fast, cited read on the evidence. For a clinician between appointments or an analyst checking a claim, that difference decides which tool gets used.
The free tier makes evaluation genuinely easy: unlimited basic searching plus a monthly allowance of AI-powered analyses, refreshing each month rather than expiring as one-time credits. You can test it properly against your own field before deciding.
Future Signal Tip: Use Consensus to decide whether a question is worth a full literature review, and Elicit to run one. The two tools cost less together than most single research subscriptions, and they solve genuinely different problems.
Consensus performs well at the job it defines for itself: returning a defensible, cited read on what published research says about a specific question, quickly.
The reliability argument is architectural rather than statistical. Because the system only summarises retrieved papers and runs checker models to verify relevance before summarising, it cannot produce the fabricated citations that have embarrassed professionals relying on general AI assistants. Every claim is clickable back to a real study. That is a meaningfully different safety property from a model that is merely accurate most of the time.
The limits are equally clear. The Consensus Meter shows how many studies agree, which is not the same as how good those studies are — a meter showing broad agreement across small, weak studies looks similar to one showing agreement across large trials. The filters help, but interpretation remains the reader’s job. And because Deep reviews are capped monthly, sustained heavy use pushes you toward the more expensive tier.
| Task | Performance | Notes |
|---|---|---|
| Quick evidence checks | Excellent | The core use case |
| Yes/no research questions | Excellent | Consensus Meter is genuinely useful |
| Health and life sciences | Excellent | Full PubMed coverage |
| Scoping a new topic | Very good | Fast orientation before deeper work |
| Systematic review | Not supported | No workflow — use Elicit |
| Your own documents | Not supported | Academic corpus only |
| Study quality judgement | Your job | The meter counts, it does not appraise |
Consensus integrates where researchers actually work. Zotero, Mendeley and EndNote are all supported for citation export, and LibKey integration lets users reach paywalled articles through their institution’s library subscriptions — a practical detail that matters more than it sounds when a promising study sits behind a paywall.
API access is available at the enterprise level, alongside institutional authentication through systems like SAML and Shibboleth, which is how universities deploy it across a campus.
What is absent is any connection to non-academic systems. There is no CRM integration, no automation platform connector, no document upload. Consensus is a research layer, not a workflow platform, and it does not pretend otherwise.
This section needs care, because most published guides are out of date.
Consensus’s own help centre now lists Pro at $20 a month or $144 a year. Numerous third-party reviews still quote figures between $8.99 and $15 a month, which reflect pricing from earlier in 2026 rather than the current rate.
| Plan | Current price | What you get |
|---|---|---|
| Free | $0 | Unlimited basic search, monthly allowance of AI analyses |
| Pro | $20/month or $144/year | Unlimited Pro messages and Study Snapshots, 15 Deep reviews monthly |
| Deep | Reported around $45/month | Everything in Pro plus 200 Deep Searches monthly |
| Teams | Custom quote | Centralised billing, individual logins, shared libraries |
| Enterprise | Custom quote | 200+ users, institutional authentication, API access |
⚠️ The price roughly doubled in 2026. Guides published in the first half of the year list Pro at around $10 a month or $120 a year. The current published figures are $20 monthly and $144 annually — an effective doubling on monthly billing and a 20% rise on annual. Annual billing is now clearly the better deal at $12 a month effective.
Two things soften this. Students and faculty with a verified academic email can get up to 40% off, which brings Pro to among the cheapest research subscriptions available. And the free tier remains genuinely usable, with monthly refreshing allowances rather than one-time credits.
Even at the new price, the value case holds for anyone doing regular evidence work. Twenty dollars against the time cost of manually searching PubMed and reading abstracts is not a close call. The question is whether you need Deep reviews often enough to justify the higher tier, and whether the 15-a-month Pro allowance fits your rhythm.
How often do you actually need to check what the evidence says, rather than what someone told you it says? For most professionals the honest answer is more often than they do.
Elicit — the tool to use when you need a full systematic review with PRISMA-compliant screening and structured extraction. The two complement rather than compete.
Scite — better when citation context matters, showing whether later papers supported or contradicted a finding.
Semantic Scholar — free discovery across an overlapping corpus, and the sensible baseline if budget is the only constraint.
Perplexity — the right choice for general web research. Consensus is deliberately worse at that, and better at academic evidence.
PubMed directly — still free and authoritative for health research. Consensus adds speed and synthesis, not access.
Consensus is a bet that the scarce thing in research is not access to papers but confidence in answers. Anyone can reach PubMed. Far fewer can quickly tell whether the evidence on a question is settled, contested or thin.
That bet is looking increasingly sound as general AI assistants get better at sounding authoritative. The fabricated-citation problem has already reached courtrooms, and professionals are learning that fluency and reliability are different properties. A tool that structurally cannot invent a source has a durable advantage that no amount of model improvement erases.
The strategic question is whether that advantage stays defensible. General assistants are adding retrieval and citation, and academic databases are adding AI layers. Consensus’s moat is the Consensus Meter and a genuinely simple experience, both of which are replicable.
The 2026 price rise is the signal worth watching. Doubling monthly pricing while keeping a generous free tier and a 40% academic discount suggests a company converting a popular free product into a sustainable business. That usually works when the paid experience is clearly better — and here it is.
Overall Rating: 8.3 / 10
Consensus does one thing extremely well: it tells you what peer-reviewed research actually says about a specific question, quickly, with sources you can click. For clinicians, analysts, students and anyone who needs to check a claim rather than assume it, that is genuinely valuable and very easy to adopt.
It loses points for the 2026 price rise, which most published guides have not reflected, and for deliberate narrowness — no systematic review workflow, no document upload, no non-academic sources.
Future Signal recommends Consensus for evidence checks, topic scoping and any professional whose decisions should rest on research rather than assertion. Pair it with Elicit if you also run structured reviews, and take the academic discount if you qualify.
How much does Consensus cost now? Consensus’s own help centre lists Pro at $20 a month or $144 a year. Many third-party guides still quote around $10 a month, which reflects earlier 2026 pricing. Annual billing is now clearly better value at an effective $12 a month.
Is there a free plan? Yes, and it is permanent rather than a trial. It includes unlimited basic searching across the corpus plus a monthly allowance of AI-powered analyses that refreshes each month rather than expiring as one-time credits.
Can Consensus make up citations? No, and this is its central design feature. It only summarises papers it has actually retrieved, with checker models verifying relevance first. It has no capacity to generate a source that does not exist, unlike general AI assistants.
What is the Consensus Meter? For yes/no research questions, it shows how retrieved studies distribute across supporting, opposing and mixed findings. It counts agreement rather than appraising study quality, so it is a starting point for judgement rather than a substitute for it.
Consensus or Elicit — which should I use? Different jobs. Consensus is for fast evidence checks and topic scoping. Elicit is for systematic reviews with screening and structured extraction. Many researchers use both, and together they still cost less than most single research subscriptions.
Does Consensus give me full-text access? Not always. It provides AI summaries and metadata, with LibKey integration to reach paywalled articles through institutional library subscriptions. Full-text availability depends on publisher licensing and your institutional access.
Is there a student discount? Yes. Students and faculty with a verified academic email can receive up to 40% off, which makes it one of the cheapest credible research tools available.
Can I use it for market or competitive research? No. The corpus is peer-reviewed academic literature and clinical research only. For news, industry reports or web content you need a different tool.
Consensus is the clearest example in this series of a tool that succeeds by refusing to do more. It will not write your paper, run your systematic review, read your documents or search the web. What it does is answer a research question from real published studies, show you how much agreement exists, and link every claim to something you can open and check.
Its weaknesses follow directly from that narrowness, plus a 2026 price rise that roughly doubled monthly cost and that most published reviews have not caught. Neither is disqualifying, but both are worth knowing before you subscribe on the basis of an outdated guide.
For clinicians, analysts, students and any professional whose credibility depends on being right rather than confident, this is an easy tool to recommend and an easier one to adopt. Start on the free tier, check the current price directly, and take the academic discount if you qualify.
Consensus
An AI search engine that answers research questions from peer-reviewed papers, and cannot invent a citation.
[Leave blank — uploaded manually in WordPress]
[Leave blank — uploaded manually in WordPress]
Consensus is an AI-powered academic search engine that answers research questions using over 200 million peer-reviewed papers, including the entirety of PubMed. Rather than generating answers from training data, it retrieves relevant studies and summarises only what they say, making fabricated citations structurally impossible. Its distinctive Consensus Meter shows how many studies support or oppose a yes/no research question. Built by Consensus NLP, Inc., it offers a permanent free tier, with Pro priced at $20 monthly or $144 annually and up to 40% off for verified academics.
Last Updated: 28 August 2026
Consensus answers research questions from peer-reviewed papers rather than from training data, which means it structurally cannot invent a citation. Ask whether something works and it returns a Consensus Meter showing how many studies agree, with every claim linked to a real study. It is fast, genuinely easy to use, and covers more than 200 million papers including all of PubMed. Two things buyers should know: it deliberately does less than Elicit, and its price roughly doubled during 2026 — a change most published guides have not caught up with.
Estimated reading time: 9 minutes
Consensus is an AI search engine for scientific literature, built by Consensus NLP, Inc. You ask a research question in plain language and it returns an answer synthesised from actual peer-reviewed studies, with every claim linked back to the paper it came from.
The corpus runs to more than 200 million academic documents, aggregated from Semantic Scholar, OpenAlex and Consensus’s own scholarly crawl. Coverage includes nearly all high-impact journals and the entirety of PubMed, which makes it particularly strong for health and life sciences.
The architectural choice that defines it is what it will not do. Consensus does not answer from a model’s internal knowledge. It retrieves papers first, then summarises only what those papers say, with checker models verifying relevance before summarisation. A general assistant can invent a plausible-looking citation; Consensus cannot, because it has nothing to invent from.
Its signature output is the Consensus Meter. Ask a yes/no research question and it shows how the retrieved studies distribute across yes, no and possibly — a fast read on where evidence actually sits rather than what the loudest paper claims.
Not to be confused with: there is a separate company called Consensus that sells demo automation software, with prices in the hundreds of dollars per month. If you see figures like $600 or $1,250 a month attached to “Consensus pricing,” you are looking at the wrong product. This review covers consensus.app, the research tool.
Consensus is a strong fit for:
Consensus is probably not the right choice for:
Consensus Meter. The standout feature. For yes/no research questions, it displays how retrieved studies distribute across supporting, opposing and mixed findings — a quick, honest read on where evidence sits.
Pro messages. AI-generated summaries drawing on up to 20 papers from a search, unlimited on paid plans. This is the everyday workhorse of the product.
Deep reviews. A more thorough analysis across up to 50 papers, aimed at genuinely exploring a complex question. Allocated monthly rather than unlimited: 15 a month on Pro, 200 on the Deep plan.
Study Snapshots. One click on any paper shows key details — methodology, sample size, study duration — without opening the full text. Unlimited on paid plans.
Ask Paper. Interrogate an individual paper directly with questions about methods, findings or implications.
Advanced filters. Narrow by publication date, study design, journal quality and sample size, which is what separates this from a general search box.
Reference manager integration. Works with Zotero, Mendeley and EndNote, with LibKey support for reaching paywalled articles through institutional library access.
Multilingual queries. Ask in a language other than English and still search the corpus effectively.
QUESTION
│
▼
SEARCH ────────► ✓ 200M+ papers, all of PubMed
│ ✗ no news, web, industry reports
▼
EVIDENCE READ ─► ✓ Consensus Meter, Pro messages
│ ✓ Study Snapshots, filters
▼
DEEP DIVE ─────► ✓ Deep reviews, up to 50 papers
│ (metered: 15/mo Pro, 200/mo Deep)
▼
SYSTEMATIC ────► ✗ No PRISMA workflow, no screening
REVIEW pipeline, no structured extraction
│ → use Elicit
▼
YOUR DOCS ─────► ✗ No upload of your own files[Illustration placeholder: Consensus coverage map]
This is the easiest research tool in its category, and that is the point. There is no workflow to configure, no extraction columns to write, no prompt engineering. You type a question and read an answer with the studies underneath it.
That simplicity is a deliberate trade. Elicit asks more of you and gives more back in structure. Consensus asks almost nothing and gives you a fast, cited read on the evidence. For a clinician between appointments or an analyst checking a claim, that difference decides which tool gets used.
The free tier makes evaluation genuinely easy: unlimited basic searching plus a monthly allowance of AI-powered analyses, refreshing each month rather than expiring as one-time credits. You can test it properly against your own field before deciding.
Future Signal Tip: Use Consensus to decide whether a question is worth a full literature review, and Elicit to run one. The two tools cost less together than most single research subscriptions, and they solve genuinely different problems.
Consensus performs well at the job it defines for itself: returning a defensible, cited read on what published research says about a specific question, quickly.
The reliability argument is architectural rather than statistical. Because the system only summarises retrieved papers and runs checker models to verify relevance before summarising, it cannot produce the fabricated citations that have embarrassed professionals relying on general AI assistants. Every claim is clickable back to a real study. That is a meaningfully different safety property from a model that is merely accurate most of the time.
The limits are equally clear. The Consensus Meter shows how many studies agree, which is not the same as how good those studies are — a meter showing broad agreement across small, weak studies looks similar to one showing agreement across large trials. The filters help, but interpretation remains the reader’s job. And because Deep reviews are capped monthly, sustained heavy use pushes you toward the more expensive tier.
| Task | Performance | Notes |
|---|---|---|
| Quick evidence checks | Excellent | The core use case |
| Yes/no research questions | Excellent | Consensus Meter is genuinely useful |
| Health and life sciences | Excellent | Full PubMed coverage |
| Scoping a new topic | Very good | Fast orientation before deeper work |
| Systematic review | Not supported | No workflow — use Elicit |
| Your own documents | Not supported | Academic corpus only |
| Study quality judgement | Your job | The meter counts, it does not appraise |
Consensus integrates where researchers actually work. Zotero, Mendeley and EndNote are all supported for citation export, and LibKey integration lets users reach paywalled articles through their institution’s library subscriptions — a practical detail that matters more than it sounds when a promising study sits behind a paywall.
API access is available at the enterprise level, alongside institutional authentication through systems like SAML and Shibboleth, which is how universities deploy it across a campus.
What is absent is any connection to non-academic systems. There is no CRM integration, no automation platform connector, no document upload. Consensus is a research layer, not a workflow platform, and it does not pretend otherwise.
This section needs care, because most published guides are out of date.
Consensus’s own help centre now lists Pro at $20 a month or $144 a year. Numerous third-party reviews still quote figures between $8.99 and $15 a month, which reflect pricing from earlier in 2026 rather than the current rate.
| Plan | Current price | What you get |
|---|---|---|
| Free | $0 | Unlimited basic search, monthly allowance of AI analyses |
| Pro | $20/month or $144/year | Unlimited Pro messages and Study Snapshots, 15 Deep reviews monthly |
| Deep | Reported around $45/month | Everything in Pro plus 200 Deep Searches monthly |
| Teams | Custom quote | Centralised billing, individual logins, shared libraries |
| Enterprise | Custom quote | 200+ users, institutional authentication, API access |
⚠️ The price roughly doubled in 2026. Guides published in the first half of the year list Pro at around $10 a month or $120 a year. The current published figures are $20 monthly and $144 annually — an effective doubling on monthly billing and a 20% rise on annual. Annual billing is now clearly the better deal at $12 a month effective.
Two things soften this. Students and faculty with a verified academic email can get up to 40% off, which brings Pro to among the cheapest research subscriptions available. And the free tier remains genuinely usable, with monthly refreshing allowances rather than one-time credits.
Even at the new price, the value case holds for anyone doing regular evidence work. Twenty dollars against the time cost of manually searching PubMed and reading abstracts is not a close call. The question is whether you need Deep reviews often enough to justify the higher tier, and whether the 15-a-month Pro allowance fits your rhythm.
How often do you actually need to check what the evidence says, rather than what someone told you it says? For most professionals the honest answer is more often than they do.
Elicit — the tool to use when you need a full systematic review with PRISMA-compliant screening and structured extraction. The two complement rather than compete.
Scite — better when citation context matters, showing whether later papers supported or contradicted a finding.
Semantic Scholar — free discovery across an overlapping corpus, and the sensible baseline if budget is the only constraint.
Perplexity — the right choice for general web research. Consensus is deliberately worse at that, and better at academic evidence.
PubMed directly — still free and authoritative for health research. Consensus adds speed and synthesis, not access.
Consensus is a bet that the scarce thing in research is not access to papers but confidence in answers. Anyone can reach PubMed. Far fewer can quickly tell whether the evidence on a question is settled, contested or thin.
That bet is looking increasingly sound as general AI assistants get better at sounding authoritative. The fabricated-citation problem has already reached courtrooms, and professionals are learning that fluency and reliability are different properties. A tool that structurally cannot invent a source has a durable advantage that no amount of model improvement erases.
The strategic question is whether that advantage stays defensible. General assistants are adding retrieval and citation, and academic databases are adding AI layers. Consensus’s moat is the Consensus Meter and a genuinely simple experience, both of which are replicable.
The 2026 price rise is the signal worth watching. Doubling monthly pricing while keeping a generous free tier and a 40% academic discount suggests a company converting a popular free product into a sustainable business. That usually works when the paid experience is clearly better — and here it is.
Overall Rating: 8.3 / 10
Consensus does one thing extremely well: it tells you what peer-reviewed research actually says about a specific question, quickly, with sources you can click. For clinicians, analysts, students and anyone who needs to check a claim rather than assume it, that is genuinely valuable and very easy to adopt.
It loses points for the 2026 price rise, which most published guides have not reflected, and for deliberate narrowness — no systematic review workflow, no document upload, no non-academic sources.
Future Signal recommends Consensus for evidence checks, topic scoping and any professional whose decisions should rest on research rather than assertion. Pair it with Elicit if you also run structured reviews, and take the academic discount if you qualify.
How much does Consensus cost now? Consensus’s own help centre lists Pro at $20 a month or $144 a year. Many third-party guides still quote around $10 a month, which reflects earlier 2026 pricing. Annual billing is now clearly better value at an effective $12 a month.
Is there a free plan? Yes, and it is permanent rather than a trial. It includes unlimited basic searching across the corpus plus a monthly allowance of AI-powered analyses that refreshes each month rather than expiring as one-time credits.
Can Consensus make up citations? No, and this is its central design feature. It only summarises papers it has actually retrieved, with checker models verifying relevance first. It has no capacity to generate a source that does not exist, unlike general AI assistants.
What is the Consensus Meter? For yes/no research questions, it shows how retrieved studies distribute across supporting, opposing and mixed findings. It counts agreement rather than appraising study quality, so it is a starting point for judgement rather than a substitute for it.
Consensus or Elicit — which should I use? Different jobs. Consensus is for fast evidence checks and topic scoping. Elicit is for systematic reviews with screening and structured extraction. Many researchers use both, and together they still cost less than most single research subscriptions.
Does Consensus give me full-text access? Not always. It provides AI summaries and metadata, with LibKey integration to reach paywalled articles through institutional library subscriptions. Full-text availability depends on publisher licensing and your institutional access.
Is there a student discount? Yes. Students and faculty with a verified academic email can receive up to 40% off, which makes it one of the cheapest credible research tools available.
Can I use it for market or competitive research? No. The corpus is peer-reviewed academic literature and clinical research only. For news, industry reports or web content you need a different tool.
Consensus is the clearest example in this series of a tool that succeeds by refusing to do more. It will not write your paper, run your systematic review, read your documents or search the web. What it does is answer a research question from real published studies, show you how much agreement exists, and link every claim to something you can open and check.
Its weaknesses follow directly from that narrowness, plus a 2026 price rise that roughly doubled monthly cost and that most published reviews have not caught. Neither is disqualifying, but both are worth knowing before you subscribe on the basis of an outdated guide.
For clinicians, analysts, students and any professional whose credibility depends on being right rather than confident, this is an easy tool to recommend and an easier one to adopt. Start on the free tier, check the current price directly, and take the academic discount if you qualify.
Our final assessment after evaluating features, performance, value, and business impact:
Consensus earns a strong recommendation for anyone who needs to check what research says rather than assume it. Its architecture — retrieving papers first and summarising only what they contain — makes fabricated citations structurally impossible, which is a meaningfully different safety property from a general assistant that is merely usually accurate. The Consensus Meter is genuinely useful, and the tool is the easiest in its category to adopt.
The caveats are worth stating plainly. Pricing roughly doubled during 2026 to $20 monthly or $144 annually, and most published guides still quote the old figures. It is also deliberately narrow, with no systematic review workflow, no document upload and no non-academic sources. Pair it with Elicit if you need structured reviews.
We may earn a commission if you purchase through our links, at no extra cost to you. This helps support independent reviews.
Explore other highly rated AI tools we’ve reviewed in this category.
