Manuscript GraderLJX Services
Chemistry · Materials · Experimental physics

A calibrated second opinion on where your manuscript stands.

Three independent language-model graders read your paper several times each, blind to who wrote it. You get one score with its uncertainty, placed against 28 journals whose standards we measured on published work — and told which of those journals actually fit your subject.

  • 3 gradersindependent models, 12–24 readings
  • 28 journalsNature and Science to Small
  • 560 paperspublished, on one fixed scale
  • £1.00per run, charged only on delivery

Usually under five minutes after grading starts. No account needed — your passphrase finds your results again.

Built for research papers in chemistry, materials and experimental physics. Other subjects receive a score, but the comparison may not be useful. Recommendations cover only the 28 listed journals.

Start a run

Prepare your manuscript

  • Keep: title, abstract, full main text, figure captions, tables and references.
  • Remove: author details, acknowledgements, funding and other back matter, journal identifiers and supporting information.
  • Text is graded. Figure images are not read. Keep captions and use editable tables.
Full preparation and file guidance

Keep the complete main paper

Include section headings in their original order, with Methods wherever you wrote it. Keep figure captions, tables as text rather than images, and the complete reference list.

Remove before uploading

  • Author names, affiliations, emails, ORCIDs and corresponding-author details.
  • Acknowledgements, funding, author contributions, competing interests, and data or code availability statements.
  • Journal names, manuscript numbers, submission history, running headers and footers.
  • Supporting information, cover letters and responses to reviewers.

Resolve tracked changes and delete comments. Finish placeholder sections and references: the graders judge the text you submit.

Choose a readable file

DOCX is preferred for headings and tables. For PDF, export a text-based, unprotected copy from Word or LaTeX; upload the compiled PDF, not the .tex source. Scanned PDFs cannot be read.

For TXT or Markdown, leave a blank line between paragraphs and put headings on their own lines.

Accepted: 2,000–20,000 words in English. The benchmark covers 3,000–12,000 words; results outside that range need more caution.

We tidy wrapped lines and heading formatting before grading. A double-blind submission copy is a useful starting point; check it against the list above.

DOCX preferred · PDF, TXT or Markdown · up to 25 MB.
2,000–20,000 words, in English.

Choose a file or drop it here

One manuscript per run

At least 12 characters; capitals matter. Reuse the same passphrase to keep your runs together in My runs. Save it somewhere safe: we cannot recover it. Anyone with it can open your runs.

Your text goes to our model providers. Your original file is discarded after extraction; manuscript text is deleted when the run ends. Results are kept for 12 months. By uploading you accept the terms and privacy note.

We check the file first. At checkout, Stripe reserves £1.00 on your card; it is charged only when your result is ready.

How it works

Four steps, no account

  1. Prepare and upload

    Use the checklist above the file picker. We check that your manuscript can be read before checkout.

  2. Reserve £1 at checkout

    Stripe authorises your card. Grading starts after checkout; the charge is taken only once a result is ready.

  3. We grade and compare

    Gemini, GLM and Kimi read the manuscript repeatedly. We calibrate their scores and assess journal fit.

  4. Read the result

    A score with its uncertainty, four groups of journals, what the graders noticed — and a link that stays readable for 12 months.

What you get

A position, not a verdict

A score with its uncertainty

One number on a fixed scale, with the spread across all the graders' readings — because a few points either side of any boundary is not a real difference.

Four groups of journals

Overkill, safe, stretch and challenge — the journals nearest your score on either side, drawn from those whose scope fits your manuscript plus the multidisciplinary titles, each with a rough chance of being sent out for review.

What the graders noticed

The strengths and doubts that recurred across their readings, in a short paragraph you can act on before submitting.

The paper as we read it

A summary of what your manuscript claims, so you can check how its scope was understood.

The 28 journals

Ordered by how demanding they proved to be

Most demanding first, measured on 560 published papers — twenty from each journal. We publish the order, never the numbers, and recommend only within this list.

  1. Nature
  2. Science
  3. Nature Physics
  4. Nature Materials
  5. Nature Nanotechnology
  6. Nature Chemistry
  7. Nature Energy
  8. Nature Electronics
  9. Physical Review X
  10. Science Advances
  11. Nature Communications
  12. Chem
  13. Joule
  14. Physical Review Letters
  15. Proceedings of the National Academy of Sciences (PNAS)
  16. Journal of the American Chemical Society
  17. Angewandte Chemie International Edition
  18. ACS Central Science
  19. Advanced Materials
  20. CCS Chemistry
  21. Advanced Functional Materials
  22. Chemical Science
  23. JACS Au
  24. Physical Review B
  25. Small
  26. Macromolecules
  27. Journal of Materials Chemistry A
  28. Chemical Communications

What this is not. It is not peer review, and the review-chance figures are a rough orientation rather than a prediction of any editor's decision — we have never seen one. It is a calibrated second opinion produced by language models that can be wrong and can be gamed. Use it to sanity-check where you were going to submit — not to decide whether the work is any good.