qvib.pro
RU

Consensus

What for: a science answer engine that gives a verdict by weight of evidence from peer-reviewed papers, with citations you can check.

платно бесплатно любой

$30M Series A (май 2026), 10M+ пользователей, 220M+ рецензируемых статей; от Allen Institute for AI (пресса) checked 2026-06-01

Updated: 02.07.2026

Open source ↗

Consensus

What it is and who it's for

Consensus is a science answer engine: instead of a list of links it gives you a conclusion grounded in peer-reviewed papers, and keeps the synthesis tied to its primary sources. Built for clinicians, analysts, PMs and anyone who needs to quickly see "what the science says" on a yes/no question with proof, rather than read a hundred papers. Its signature feature is the Consensus Meter, which boils the weight of evidence down to a visual "yes / no / possibly" in seconds across a corpus of 220M+ works.

Key features

  • Answers to research questions grounded in peer-reviewed papers, with inline citations.
  • Consensus Meter — a visual "yes / no / possibly" summary of the weight of evidence.
  • A corpus of 220M+ papers (OpenAlex + Semantic Scholar + exclusive full-text partnerships).
  • Per-paper summaries and filters by study type.
  • Integration with GPT models; library subscriptions at 170+ universities.
  • Export and saving of findings.

Getting started in 5 minutes

  1. Open consensus.app and ask a question like "Does X help with Y?".
  2. Look at the Consensus Meter and click through to the source papers to verify.
  3. For depth, narrow down by study type (meta-analyses / RCTs) and save the collection.

When to use it and when not to

  • ✅ Use it if you need a fast evidence-based "yes / no / possibly" backed by peer-reviewed proof.
  • ✅ Use it if speed and a clear read on the weight of evidence matter more than a full systematic review.
  • ❌ Skip it → go with "Elicit" if you need a structured literature review with data extracted into tables.
  • ❌ Skip it → go with "Perplexity" if the question isn't scientific but general web research.

Honest pricing

Free plan with a query limit; paid plans add more queries, stronger models and advanced filters. Often available through university libraries. Exact numbers are on the vendor's site and do change.

Gotchas

  • "Yes / no / possibly" is a simplification: for clinical or high-stakes decisions, read the papers and their context yourself.
  • Answer quality depends on how you phrase the question (binary, specific questions work best).
  • Coverage is uneven across disciplines; paid full text isn't available for every journal.
  • It doesn't replace a systematic review — it's a tool for quick evidence-based orientation.

🤖 Prompt booster

"I have a question <DESCRIBE — ideally binary and specific> for Consensus. Help me rephrase it so the Consensus Meter returns a clear weight of evidence; suggest which study-type filters to turn on (meta-analyses / RCTs), and give me a checklist of which 2-3 key papers to open and what to check in them before I rely on the conclusion."

Читать по-русски →