Lens API 与 SciSpace API 对比：专利与论文检索能力差异

For a product that answers questions across patents and papers, the SciSpace API is the better fit: one plain-language research task covers discovery, reading, extraction and cited writing, and returns finished files. Choose the Lens API when your main deliverable is structured patent records or explicit patent-to-paper citation relationships. Lens documents those record fields directly.

SciSpace searches its 280M+ paper index plus Google Scholar, PubMed and arXiv, and a task can also search Google Patents, clinical-trial registries, grant databases and 77 public science databases. It does not query Lens, and it is not a replacement for Lens's patent-record schema.

Lens API vs SciSpace API: what changes for your product?

The central difference is the output your application consumes. Lens returns search records; SciSpace's Task API runs a research task and returns links to the files the agent created plus any supporting text answer. Lens's search response includes total and data.

Lens's documented resources and schemas support the record-oriented side of this comparison. The additional synthesis work is an architectural implication, not a claim about every Lens website feature.

What do the same five research questions produce?

Use these questions as an acceptance suite. Each asks which papers and patents describe a technology and what evidence supports its performance.

Method: these are documentation-based workflow comparisons, not authenticated test results. The Lens column describes retrieved inputs and application work; the SciSpace column describes the requested deliverable, not a verified scientific answer.

Lens supplies queryable records through its scholarly and patent resources; the extraction steps above would belong to your application.

Compare retrieval relevance separately from synthesis quality. Check whether measurements have units and conditions, unavailable evidence is marked, and citations actually support each conclusion. Topic similarity must not be presented as an explicit patent-to-paper citation link.

Our patent search API guide explores the combined discovery use case.

How do you send the questions to Lens?

After Lens approves access, its POST endpoints accept bearer authentication. This shortened Python example preserves the original two-resource pattern: one scholarly search and one patent search for the first question.

The query and projections follow the separate request and response schemas. Replace the topic expression for each remaining question, keeping the full research question in your evaluation record.

This retrieves only an initial page. Production code should implement pagination, appropriate retry handling and query logging before anyone interprets a small result sample as a coverage test. Lens documents pagination controls and retry headers.

What does a SciSpace request look like?

You send the same first research question to the Task API in plain language, plus instructions for evidence handling. For example: "Which papers and patents describe solid electrolytes for lithium batteries, and what evidence supports their conductivity claims? Compare measurements and conditions, cite supporting papers, separate patent disclosures from experiments and mark missing evidence."

The agent picks the tools: searching papers and Google Patents, chasing citations forwards and backwards, reading full text, extracting and writing. The plan and progress stream live while it runs. What comes back is links to every file the agent created, such as a Markdown or Word report and an extraction table, plus any supporting text answer. Follow-ups continue in the same thread.

What comes back, side by side?

These are response contracts, not captured answers to the five questions. Lens's documented envelope supplies the record-oriented fields below.

Lens's scholarly citation fields represent document relationships, not evidence annotations attached to a generated answer.

For SciSpace, an extraction table is one of the files a task can return. AI-filled columns from full text can capture conductivity, method or test conditions, and you can then filter, sort and combine.

Keep patent identifiers separate from paper citations. SciSpace's patent search runs through Google Patents and does not return Lens-style structured patent-family or legal-status records.

For document-centered research, SciSpace can answer questions across up to 10 PDFs at once with page-level citations, and it reads figures and tables.

Lens API pricing, evaluation access and operating limits

Checked October 31, 2026: Lens specifies custom API pricing rather than a universal dollar amount per 1,000 calls. Its terms say API access requires an institutional subscription and that trial access does not include commercial permissions.

The trial description is not a universal ongoing free allowance. Confirm eligibility with Lens, particularly where the support page and institutional-access terms require clarification.

Use returned quota headers rather than hard-coding an assumed allowance. For workload estimates, remember that the example performs two initial Lens searches per question before pagination or downstream evidence processing.

When should you choose Lens?

Choose Lens when structured patent data is the product, not merely an input to an explanation. Its patent schema includes families, applicants, classifications, claims and references; its scholarly schema includes patent citation identifiers.

Good reasons to retain it include:

• Users need explicit patent-family or citation relationships.
• Your product depends on inspectable query predicates and record projections.
• Your team already maintains the reading and synthesis pipeline.
• You need a record browser rather than a research assistant.

Lens documents field selection and query controls for this style of integration.

Review the Lens API terms and your agreement before planning customer-facing reuse. Access approval and permitted reuse are separate checks.

How do you map Lens parameters to SciSpace?

Switching changes the application contract, not just the endpoint. Lens's query, projection and pagination controls do not translate directly into research instructions.

Plan the frontend around the live stream of plan steps and progress, then the returned files. Preserve original identifiers, display citations beside claims, and validate hard constraints independently. Follow-ups continue in the same thread.

Why the SciSpace API is a better fit for patent-and-paper research tools

Lens stops at records: families, parties, classifications, claims and citation links. Reading the evidence behind a technology, comparing measurements across papers and patents and writing a cited report are layers you build on top. Lens remains the better record backend when patent schema fidelity is the requirement.

The SciSpace API puts the full SciSpace Agent behind one call. It searches, reads, analyses and writes, and delivers the result as ready-to-use files.

What this adds on top of Lens

• Evidence beyond papers. One task can search 280M+ papers and also 11 national and regional clinical-trial registries plus the WHO registry, Google Patents, grant databases (Grants.gov, foundations, international funders) and 77 public science databases such as PubChem, ChEMBL, UniProt, ClinVar and PDB.
• One integration instead of several. Describe the task in plain language and the agent picks the tools: search, screening, extraction, analysis and writing. You don't stitch together a search API, a PDF parser and an LLM.
• Extraction tables from full text. AI-filled columns from full text, which you can then filter, sort and combine.
• Finished files from one run. Markdown, Word, PowerPoint, or a LaTeX report compiled to PDF with its BibTeX, plus plots, posters and scientific illustrations when the task calls for them.
• Live streaming and follow-ups. The plan and each step stream while the task runs, the answer streams as it is written, and follow-ups continue in the same thread.

When to use which

• Use Lens when structured patent records, patent families or explicit patent-to-paper citation relationships are the product.
• Use the SciSpace API when your product explains the evidence across patents and papers and delivers cited reports and tables.

What it doesn't do

• No Lens-style patent schema. It does not query Lens, and patent search runs through Google Patents, without Lens's family, claims or legal-status records. It also does not query Semantic Scholar, Scopus or Web of Science.
• No paywalled papers. It reads your own PDFs plus 50M+ open-access full texts.
• Not a web search API. For news, product docs or general web content, use a web search API.

The SciSpace API is in early access; it isn't open yet. At launch it's open to anyone: register on SciSpace and start, with no sales call. Every account gets $5 in free API credits, enough to try both APIs on real questions. The API is separate from SciSpace consumer plans; no consumer plan includes API usage.

Frequently asked questions

Last checked

October 31, 2026. Lens facts were checked against official documentation. SciSpace details reflect the API's early-access status on the same date. No authenticated accuracy or performance benchmark was run.

Sources

1. Lens API Terms of Use
2. Lens API Documentation
3. Lens API Getting Started
4. Lens Scholar Request
5. Lens Patent Request
6. Lens API Response
7. Lens Scholar Response
8. Lens Patent Response

注意转载授权 · 阅读原文