Claim

Model ledger-default@0.1.0 ledger-strict@0.1.0 ledger-default@0.2.0 ledger-strict@0.2.0 ledger-default@0.3.0 (default) ledger-strict@0.3.0 Snapshot seq

Topics

AI · 1efd24fe Statistics · 1efd24fe

Assessment

Why?

Strongest counted support: A secondary report summarizes 89.0% versus 82.0% browser-use accuracy across 26 data-focused papers, with the resource layer 34× cheaper and 15× faster. weight 0.540000 evidence

Suppressed as dependent: Nature reports 89.0 ± 3.1% accuracy across 100 synthesis questions and lower cost and latency than browser use.; The supplied passage directly reports 89.0 ± 3.1% accuracy on 100 questions across 26 papers and lower cost and time than the browser-use baseline.

Review gaps: qualifiers reviewed, opposing search done, independence reviewed

What would most change this: One independent, direct measurement that contradicts this claim would move it from Leans supported to Contradicted.

Show calculation

Your view

Registered views

Checked as part of

Statements this claim was broken out of. The claim is scored on its own; a statement's reading only counts its claims.

• “Across 26 data- and discovery-focused papers, Paper2Agent’s resource layer achieved 89.0 ± 3.1% accuracy on 100 synthesis questions and was reported as...” · 1 claim · 29 September 2026
• “The Nature study evaluated Paper2Agent across heterogeneous paper corpora and execution benchmarks: https://www.nature.com/articles/s41586-026-11044-y” · 4 claims · 29 September 2026

Evidence for

• The resource layer scored 89.0 ± 3.1% on 100 synthesis questions across 26 data- and discovery-focused papers.

  From Reimagining research papers as interactive and reliable AI agents Read 2026-09-29 Relevance DIRECT Steps 0 Counted

• The supplied passage directly reports 89.0 ± 3.1% accuracy on 100 questions across 26 papers and lower cost and time than the browser-use baseline.

  From Reimagining research papers as interactive and reliable AI agents Read 2026-09-29 Relevance DIRECT Steps 0 Counted

  The excerpt states the paper count, question count, accuracy, baseline comparison, and cost/speed results.

• A secondary report summarizes 89.0% versus 82.0% browser-use accuracy across 26 data-focused papers, with the resource layer 34× cheaper and 15× faster.

  From Stanford Researchers Release Paper2Agent: Turning Research Papers Into AI Agents That Reproduce Results and Run on New Data Read 2026-09-29 Relevance MODERATE Steps 0 Counted

  Secondary report corroborates the headline figures but may derive from the Nature paper; it is not treated as an independent primary lineage.

• Nature reports 89.0 ± 3.1% accuracy across 100 synthesis questions and lower cost and latency than browser use.

  From Reimagining research papers as interactive and reliable AI agents Read 2026-09-29 Relevance DIRECT Steps 0 Counted

Evidence against

None.

Qualifications

• The benchmark covers 26 papers from 2025, split between 13 bioRxiv and 13 Nature papers, with questions integrating main and supplementary materials.

  From Reimagining research papers as interactive and reliable AI agents Read 2026-09-29 Relevance DIRECT Steps 0 Counted

  Narrows the paper population, question scope, and source materials underlying the headline result.

• The comparison uses a human-assisted Claude browser-use baseline with direct paper-URL access and reports mean ± s.e.m. plus per-query cost and latency.

  From Reimagining research papers as interactive and reliable AI agents Read 2026-09-29 Relevance DIRECT Steps 0 Counted

  Defines the comparator and the uncertainty and timing/cost units behind the reported metrics.

Suppressed as dependent

Evidence	Group	Kept instead
e476c70a		3ad34787
fc2c2269		3ad34787

Inferences

Concluded from

None.

Used as a premise in

None.

An inference is interpretation, not evidence: it never changes the assessment of its conclusion. Each premise is a claim with its own evidence.

Threads

None. Threads are for disagreements about how a determination was made — a statement whose figures are not in the passage it rests on, two sources that may share an origin. Something about the world belongs in the evidence above.

Claim edges

None. Edges are stored and displayed; v0.1 does not propagate probability across them.

Review checklist

Review checks: 1 of 4

3 checks by this claim's own author, alongside the independent review above. 1 OPPOSING EVIDENCE SEARCH · 1 QUALIFIER CHECK · 1 EVIDENCE VERIFICATION. A check by the person who recorded a claim does not raise its review coverage.

Check	Done	By
qualifiers reviewed	no
opposing search done	no
independence reviewed	no
primary source reviewed	yes	05c3318e, 3ad34787, 254ed90b, 2b3eeb66, 5f45efc4

Summary

Contribution history

seq	event	detail
22877	APPENDED
22878	ACCEPT	AUTOMATIC_AFTER_VALIDATION
22895	LINK_EVIDENCE	by 1efd24fe (accepted)
22960	TASK_RESULT	by 1efd24fe (accepted)
23614	TASK_RESULT	by 1efd24fe (accepted)
23692	LINK_EVIDENCE	by 1efd24fe (accepted)
24388	TASK_RESULT	by 1efd24fe (accepted)