Models
A mixture of models, cross-checked.
Every frontier model (Claude, GPT, Gemini, Mistral, Grok) alongside our own, routed per sub-task and made to check each other. No single-model bias.
Explore Science vs q.e.d. Science
q.e.d. Science maps a life-sciences paper or grant into a claim tree and returns a rigor report in about thirty minutes. Explore Science is the AI co-scientist for any field: a longer cross-checked review with a Calibre score, every reference verified, and a chat that has read your work.
The short answer
Choose Explore Science, the AI co-scientist, for a review that applies one rubric to any field, verifies every reference, and tracks how each revision moved its Calibre. Choose q.e.d. Science for a fast claim-structure pass on a life-sciences manuscript or grant, especially if you submit through bioRxiv and hold an institutional email. Both go past surface comments into the reasoning of a paper. The difference is that q.e.d. Science bets on one method tuned to one field, while Explore Science spends up to two hours letting frontier models and scientific agents check each other across whatever field you work in.
Compared July 2026 · from q.e.d. Science's own published materials
Choose Explore Science when
- Field-neutral: the same eight-category rubric applies to physics, chemistry, or the social sciences, not the life sciences alone.
- Every reference is verified against live scholarly databases, flagging any that does not resolve or does not support the claim attached to it.
- Claude, GPT, Gemini, Mistral, and Grok run alongside our in-house Explorer One and reconcile their findings; the q.e.d. Science engine is not disclosed.
- Every revision is scored and diffed against the last, inside a shared workspace your co-authors can join.
- Rosa, the co-scientist chat, answers on your manuscript, its review, and the rest of your published work.
Choose q.e.d. Science when
- Faster: a full report in about thirty minutes, against our two-hour ceiling.
- Claim-tree decomposition is a distinct method, mapping how each claim leans on its evidence, and it runs on grant proposals as well as manuscripts.
- An official bioRxiv on-ramp routes a preprint straight into review, which Explore Science does not have.
- QED Score is published and validated at scale: 57,455 bioRxiv preprints were scored to build its 1% ranking in June 2026.
- Named endorsements from professors at Harvard, Yale, Oxford, Tufts, and Berkeley sit on its homepage; we publish nothing comparable yet.
Side by side
| Capability |  | q.e.d. Science |
|---|---|---|
| What it reviews | Manuscripts in any field, plus protocol/proposal, editorial, and novelty reviews | Manuscripts and grant proposals; its published work and benchmarks are life sciences |
| TurnaroundPer full report | Up to two hours for a full pass; emailed when it lands | About thirty minutes |
| Review engine | Claude, GPT, Gemini, Mistral, and Grok run against each other alongside our own Explorer One, then reconcile | Not disclosed |
| Scoring | Calibre: a score out of 100 across eight categories, with sub-scores | Claim-tree decomposition, plus QED Score: originality and validity, benchmarked against similar papers |
| Reference checking | Every reference checked against live scholarly databases | Not described in its published materials |
| Revisions | Every version rescored on the same rubric, with the movement shown | No revision-scoring workflow described |
| Paper-grounded chat | Rosa answers on your manuscript, its review, and the rest of your published work | None described for the researcher product; a separate enterprise Validity Machine, in early access, answers questions across the literature |
| bioRxiv on-ramp | None | Official integration to route a preprint into review |
| Access | Open to everyone; no waitlist, no institutional email | Requires an institutional or university email; access link sent immediately, no waitlist |
| Pricing | $49 for one full review with no subscription, or flat plans from $99 a month with an allowance covering reviews and chat | Free for academic researchers with an institutional email; a paid enterprise tier for pharma and biotech, price not published |
| Training on your work | Does not train models on your manuscript; you can delete it | States it does not train on uploaded manuscripts or grants; its terms keep an optional licence to train and evaluate QED's own models, and its FAQ notes a manuscript can indirectly affect model parameters |
90% rate a review on par with or better than the human peer review they have had.
Two in three rate it better. From 304 head-to-head ratings by researchers who had received peer review on the same paper. These are our own users, self-reported, which is worth knowing when you weigh them.
Depth
Why a thorough review takes the time it takes.
Most tools optimize for speed. Explore Science runs the engine we built to conduct original research autonomously, and points it at your manuscript: it reads recent literature in your field, checks every reference, and works through the paper the way a careful reviewer does.
The time is the method, not an overhead on it.
How a paper is scoredTime spent thinking about your paper
How long a review actually takes
Single-pass LLM
~2 min
Other AI reviewers
5 to 20 min
Human peer review
(active reading time)
~30 min
Explore Science
up to 2 hrs
Industry observation; self-reported vendor figures; Publons peer-review time survey, 2018.
The engine
What actually reads your paper.
Three pieces of the architecture do the work behind every review.
Agents
Scientific agents that reason like the field.
Dozens of specialised agents work each review in parallel, each tuned to one dimension of analysis, spawning more when a section needs them.
Infrastructure
An in-house science stack.
Memory that holds your paper in context for the whole review, live verification of every citation, and novelty assessment. Built over years of autonomous science.
Claim trees do one job, and do it fast
q.e.d. Science breaks a manuscript into its constituent claims, exposes the logical and evidential links between them, and benchmarks the result against similar papers using its QED Score, which grades originality and validity. It ships that in about thirty minutes, it runs on grant proposals too, and it integrates with bioRxiv. For a life-sciences author who wants the argument stress-tested before reviewers see it, that is a strong offer.
Explore Science makes a different bet. Instead of one method tuned to one field, it spends up to two hours letting frontier models and scientific agents check each other, then scores against the same eight-category rubric whether the paper is molecular biology or condensed-matter physics.
Why we verify the bibliography
Explore Science checks every reference against live scholarly databases and flags any that does not resolve, or that does not support the claim attached to it. As more drafts pass through generative tools, a bibliography that is formatted perfectly but points at the wrong paper has become a common and costly error.
q.e.d. Science describes no reference or citation verification in its published materials, and it states that it assumes the underlying data and results are genuine; its stated purpose is to interrogate the reasoning, not to detect manipulation. If reference integrity is part of what you need checked, ask them directly before relying on it.
Common questions
Can I use q.e.d. Science without a university address?
Its free account form requires an institutional email and an academic status, so industry and independent researchers are routed to its enterprise sales team instead. Explore Science has no such gate: any address works, and a single review costs $49 with no subscription.
Will my manuscript be used to train q.e.d. Science's models?
QED states it does not train on uploaded manuscripts or grants, and that its AI providers are contractually barred from training their foundation models on your data. Its terms still grant it an optional licence to train and evaluate its own models, and its FAQ notes that a manuscript can indirectly affect model parameters, which it says cannot be rolled back. Deleting a project ends that licence, with backups purged within thirty days. Explore Science does not train models on your manuscript at all.
Could running a draft through either tool count as prior publication?
Neither is a publication venue. QED describes itself as a private pre-submission tool whose reports stay in your account unless you share them, and it suggests an acknowledgements line if you want to credit it. Explore Science works the same way: your manuscript and its reviews stay in your workspace until you invite someone.
I have one paper, not a subscription budget. What does each cost?
q.e.d. Science is free if you qualify for academic access, which is hard to beat on price alone; it publishes no paid researcher tier and no per-paper price. Explore Science charges $49 for one full review with no subscription, so a single paper does not commit you to a plan, and the $99 monthly tier earns its place once you are running several reviews or leaning on Rosa between drafts.
Claims about q.e.d. Science reflect its own published materials; everything about Explore Science reflects what the platform does today. Sources: q.e.d. Science homepage and FAQ, QED Science terms of use, What is QED Score? (QED Science), QED Science launches AI infrastructure for scientific validation, bioRxiv: q.e.d Science, an AI review tool for authors, The Scientist: QED, an AI tool for smarter manuscript review. Something out of date? and we will fix it.
Already using q.e.d. Science? Put the same paper through a scored review.
Get started free