If you need a fast read on scientific claims, Consensus searches and summarizes academic papers with direct links to the sources. It shows an evidence synthesis called the Consensus Meter so you can see how many retrieved studies support, oppose, or give mixed results on a claim.
You can run unlimited paper searches and open Study Snapshots to extract key details without reading every paper. There is a free tier and a Pro plan priced at $20/month, and a seven-day free trial lets you test Pro features before committing.
Who it's for
Consensus is built for people who need quick, evidence-rooted answers and direct links to the research behind a claim. You will be disappointed if you need reproducible, auditable searches or precise effect sizes for formal reviews.
Buy it if
- You are a freelance health or science writer with tight deadlines and need a source-linked summary to start drafting.
- You run a small content or PR workflow and want quick, citable summaries that point to original papers.
- You use Google Scholar or PubMed and want a faster, AI-driven orientation before you open full papers.
Skip it if
- You need searches that are auditable and fixed for registered systematic reviews or regulatory work.
- You require numeric effect sizes, confidence intervals, or risk-of-bias tables ready for meta-analysis.
- You expect every AI summary to be error-free without checking the original papers.
Your first hour
Sign up with an email address or use Google sign-in on the web or mobile app, and a 7-day trial of Premium is available. You can run searches immediately and get a usable summary in seconds.
- Create an account Sign up with email or Google on the web or mobile; follow prompts and start the trial if offered.
- Run a search Type a question and get an AI summary with direct links to source papers in seconds.
- Try a Pro Analysis Free accounts have limited Pro messages, while paid Pro unlocks unlimited GPT-4 powered Pro Analyses.
- Open a Study Snapshot Free accounts include Study Snapshots and a small number of Deep reviews per month to test deeper extraction.
The learning curve is short for quick orientation searches, but reading source papers and interpreting the Consensus Meter takes some practice.
What you actually get per plan
Consensus keeps a usable free tier and several paid subscriptions. Monthly and discounted annual billing are published for individual plans, while Teams and Enterprise require custom quotes and per-seat negotiation.
Most readers here will end up on the Pro plan because it removes message limits and unlocks more Deep reviews. Pro is published at $20/month or $144/year billed annually.
| Plan | Price | What’s inside | Catch |
|---|---|---|---|
| Free | $0/mo | Unlimited paper searches; limited Pro messages; 3 Deep reviews and 10 Study Snapshots per month | — |
| Pro | $20/month ($144/year billed annually) | Unlimited Pro messages; Unlimited Study Snapshots; 15 Deep reviews per month | Discounted rate requires annual billing |
| Deep | $65/month ($540/year billed annually) | 200 Deep reviews per month; Unlimited paper searches; 1,000 API & MCP calls per month | — |
The free plan has hard caps on Deep reviews and Study Snapshots (3 and 10 per month respectively).
Three real situations
Here are three real users and what Consensus actually does for them.
A freelance health writer with three retainer clients
They pay $20/month for Pro and get quick, source-linked summaries that cut initial research time.
A two-person online supplement store
They rely on the free tier and 10 Study Snapshots, but hit that 10-per-month cap during product launches.
A solo researcher doing a systematic review
Consensus is not the right buy for reproducible searches; use PubMed or exportable tools like Elicit instead.
Where it falls short
Reviewers say Consensus is fast for orientation but not reproducible.
It can misinterpret numeric and reasoning-heavy questions. It also produces fluent, confident-sounding summaries that can be wrong. The Consensus Meter counts papers equally and does not weight by study design or sample size. Search results are sensitive to query wording and can change on reruns. Compared with Elicit, Consensus is weaker for building structured, exportable evidence tables.
Subscriptions auto-renew and cancellations only take effect at the period end, which hurts annual plan buyers. Users have reported unexpected billing and difficulty cancelling. Restoring reduced limits reportedly costs 4.5× the subscription, which hits solo buyers who downgrade. Search output is English-first, which is a problem for non-English researchers. Refund terms are not clearly published, so support delays can hurt small teams on tight budgets. Consensus processes content with third-party models but says it does not use user data to train those models, which matters if you need strict audit trails or reproducible retrieval for formal reviews.
Watch out: Annual plans auto-renew and cancellations may leave access active until the period end, and users report unexpected billing and cancellation problems.
Common questions
Is there a free plan?
Yes. The free plan includes unlimited paper searches and limited Pro messages.
It also includes 3 Deep reviews, 10 Study Snapshots, and 30 API and MCP calls per month.
Can I cancel any time?
Yes. Cancellation takes effect at the end of your current billing period.
On annual plans access can remain until the period end, which may be up to 12 months.
Do they offer refunds?
Consensus does not publish a refund policy on its pricing page.
Contact support for refund requests and case-by-case help.
Does Consensus use my data to train models?
No. Consensus says it does not use user data to train large language models.
Content is processed by third-party models under agreements that prevent using your data for training.
What is the cheapest way to try it?
Use the free plan or start the 7-day free trial of Premium to test Pro features.
Bottom line
Buy it if you need fast, source-linked literature summaries and will use it enough to justify $20/month. Skip it if you require reproducible, auditable searches or numeric effect sizes; use PubMed or Elicit instead.
