← Back to API Docs

Frequently Asked Questions

Everything you need to know about VerifyScience API Beta

🚀 Getting Started

How long does verification take?

Typical time: 30-60 seconds per paper

The API uses asynchronous processing:

  1. You submit a request → Get instant job_id
  2. Paper queues for processing (~5-10 seconds)
  3. Claude analyzes methodology (~30-45 seconds)
  4. Results available via GET /api/verify/:id

Maximum time: 2 minutes for complex papers

ℹ️ Processing times may vary

Actual processing time depends on several factors:

  • Network speed and latency
  • Claude API response time (varies by load)
  • Paper length and complexity
  • Current queue depth

During peak times or network slowdowns, verifications may take 2-3 minutes. This is normal.

Do I need my own Claude API key?

Yes. VerifyScience uses a "bring your own key" model.

Why this model?

  • ✅ You control your AI costs directly
  • ✅ No surprise bills from us
  • ✅ We keep subscription prices low (95% margin = lower fees)
  • ✅ Your key = your Claude usage limits

How to get a key:

  1. Sign up at console.anthropic.com
  2. Navigate to API Keys section
  3. Generate new key
  4. Add it in VerifyScience Settings

Security: Your key is encrypted with AES-256-GCM before storage.

What's the difference between plans?

Feature Free Starter Professional Enterprise
Verifications/month 3 199 499 1,999
Batch verification (Professional+ only) ✅ 20/batch ✅ 50/batch
API rate limit 10/hr 50/hr 100/hr 500/hr
Price Free £4.99/mo £9.99/mo £19.99/mo

🔧 Errors & Troubleshooting

Error: QUOTA_EXCEEDED

What it means: You've used all monthly verifications

Solutions:

  1. Wait for reset: Quota resets 1st of each month at 00:00 UTC
  2. Upgrade plan: Get more verifications instantly
    • Starter: 199/month (£4.99)
    • Professional: 499/month (£9.99)

Error: INVALID_DOI

DOI format must be: 10.XXXX/identifier

✅ Valid examples:

  • 10.1038/nature12373
  • 10.1371/journal.pone.0123456
  • 10.1016/j.cell.2024.01.001

❌ Invalid examples:

  • nature12373 (missing prefix)
  • https://doi.org/10.1038/nature12373 (use DOI only)
  • 10.1038 (incomplete)

Error: RATE_LIMIT_EXCEEDED

Rate limits by tier:

Tier Requests/Hour
Free10
Starter50
Professional100
Enterprise500

How to handle:

  • Implement exponential backoff in your code
  • Upgrade tier for higher limits

🧩 Chrome Extension

How do I install the Chrome extension?

  1. Visit the Chrome Web Store listing
  2. Click "Add to Chrome"
  3. Click the puzzle piece icon in Chrome's toolbar and pin VerifyScience
  4. Click the extension icon → "Connect to VerifyScience" (one-time setup)

Which sites does the extension support?

Currently supported: PubMed (pubmed.ncbi.nlm.nih.gov)

Coming soon: bioRxiv, medRxiv, Google Scholar

On unsupported sites, the extension shows "Navigate to a Paper" with a list of supported sites.

Does the extension track my browsing?

No. The extension accesses only paper metadata (DOI, title, authors, abstract) from PubMed pages. No browsing history is tracked. No data is transmitted until you explicitly click "Verify This Paper".

Do extension verifications count against my quota?

Yes. Extension verifications use the same monthly quota as the web platform. Your quota is shared across all access methods (web, API, extension).

📋 Reference List Verification

What is reference list verification?

Upload an entire bibliography and scan every citation for retraction status, citation metrics, and open access availability. Three tiers:

  • Tier 1 — Signal Scan: Free, instant. Cross-references every citation against OpenAlex and Crossref.
  • Tier 2 — Quick Check: AI abstract triage. 1 credit per reference.
  • Tier 3 — Deep Verify: Full 12-criterion assessment. 1 credit per reference.

How many references can I scan?

Limits depend on your subscription tier:

  • Free: Reference scan not available
  • Starter: Up to 10 references per scan
  • Professional: Up to 50 references per scan
  • Enterprise: Unlimited

⚙️ Technical Details

What languages are supported?

Multi-language supported. VerifyScience can analyse papers in multiple languages. A language badge is displayed for non-English papers. Analysis quality is strongest for English and major European languages.

How accurate is verification?

Using Claude Sonnet 4 with structured prompting:

  • 85-90% agreement with human experts
  • <10% false positive rate
  • Best at: Sample size, statistical tests, design
  • Limitations: Novel methods, field norms, theory

⚠️ Use as decision support, not final judgement

Still have questions?

Our team is here to help

Response time: <24 hours (Beta testers: <4 hours)