Contents 18 sections

The best AI for research depends on the job: a cited web report, a notebook built around your sources, or a search for academic papers. ChatGPT Deep Research searches, compares, and brings web sources together in cited reports.1
This ranking compares ChatGPT, Claude, Gemini, and specialist tools by sources, citations, limits, and price. OpenAI publishes an allowance of 25 deep research queries per month for ChatGPT Plus.2

Comparison of AI research tools: table and ranking criteria

Perplexity, ChatGPT deep research, Gemini Deep Research, Gemini Notebook (NotebookLM), Claude Research, Elicit, Consensus, Scite, SciSpace, Semantic Scholar, ResearchRabbit, and Undermind address different research tasks. This ranking weighs source scope – what each tool searches or analyzes – alongside citation visibility, free access, paid price, and usage limits. Its order does not mean one tool leads on every measure.

The table compares source scope, citations, free access, monthly paid prices, and research limits in ranking order. As one price reference, ChatGPT Plus costs $20/month.3

Tool Sources searched
1. Perplexity Web
2. ChatGPT deep research Web
3. Gemini Deep Research Web
4. Gemini Notebook (NotebookLM) Your files
5. Claude Research Web and connected data
6. Elicit Academic papers
7. Consensus Academic papers
8. Scite Academic papers
9. SciSpace Academic papers
10. Semantic Scholar Academic papers
11. ResearchRabbit Academic papers
12. Undermind Academic papers

Table data: 3

– means this comparison does not give a verified value, not that a plan or feature is unavailable. Source-scope labels describe each tool’s research category, not every source it can access. The stated Consensus allowances do not identify a plan, so the table leaves them unassigned.4 Consensus Free includes unlimited Papers searches and three Deep Reviews per month.5

Source scope shapes the questions a tool can answer. Web tools suit current topics and companies; notebooks work from chosen material; literature tools focus on papers. Elicit says systematic-review researchers have reported up to 80% time savings, but that outcome is not guaranteed for every review.6 SciSpace offers a 100% refund within 24 hours of purchase only if the customer has used fewer than 300 Agent credits. That is a purchase condition, not a research-output limit.7,8

ResearchRabbit is the clearest choice for readers who want a paper-discovery tool with both free and paid tiers priced here. For cited web reports, selected files, or academic literature, start with source scope. Then check important claims against their original sources.

1

Perplexity

At a glance

  • Perplexity Pro, the lowest-priced plan offering premium finance sources, costs $20/month.
  • In a 100-task benchmark, Perplexity Deep Research scored 90.24% for citation accuracy: whether cited sources supported the related statements.

Perplexity suits finance-oriented research when premium data sources matter more than a dedicated academic-paper workflow. Its value depends on access to the right sources, citations that support the claims, and plan cost. Perplexity Pro, the lowest-priced plan offering premium finance sources, costs $20/month.9

Perplexity: features confirmed by the vendor

Perplexity points to premium data sources for finance and analysis.10 That focus suits market and company questions. Literature reviewers, by contrast, need to assess academic-paper coverage on its own terms. Perplexity names CB Insights, PitchBook, and Statista as premium market-data providers.11

Perplexity: pricing, limits and citation quality

Perplexity shares a customer account saying time and financial savings exceeded an annual license’s cost in the first week.10 It is a customer account, not a plan price or promised return. In a research report, every cited source should support the particular claim beside it.

Perplexity: who is it best for?

Perplexity is best suited to analysts who put premium finance sources ahead of a paper-centered literature workflow.

  • Finance and company analysts should compare Perplexity’s available data sources with the questions they need to answer.
  • Academic researchers should not choose Perplexity for its finance focus if finding and evaluating papers is their main task.

Premium finance sources give Perplexity its clearest place in this ranking, but its reports still need citation checks. In a 100-task benchmark, Perplexity Deep Research scored 90.24% for citation accuracy: whether cited sources supported the related statements.12

Independent, data-backed
2

ChatGPT deep research

3.7/5

At a glance

  • OpenAI says completion can take 5 to 30 minutes.
  • ChatGPT deep research typically produces a report in 5 to 30 minutes.
  • In the same 100-task benchmark, OpenAI Deep Research scored 77.96% for citation accuracy.

ChatGPT deep research suits researchers who need a cited report from web and connected sources. Perplexity, by contrast, puts more stated emphasis on premium finance data. The useful comparison is source access, report workflow, and support for each cited claim.

ChatGPT deep research: features confirmed by the vendor

ChatGPT deep research can review the web and draw on hundreds of sources for a report.1 While it runs, researchers can edit the plan, refine the goal, change sources, or stop the task.1 It can prioritize connected private files, proprietary MCP connectors, paid datasets, and approved URLs; reports include citations.1 OpenAI says completion can take 5 to 30 minutes.2

ChatGPT deep research: pricing, limits and citation quality

ChatGPT deep research typically produces a report in 5 to 30 minutes.1 Researchers can share the finished report, download a PDF or DOCX, or refine it in chat.1 ChatGPT lists context windows – how much material a model processes at once – of 27K for Free GPT Instant, 54K for Go and Plus GPT Instant, and 256K for Go GPT Reasoning. Those figures are not deep-research report allowances.13 Important claims still call for a check against the original source.14

ChatGPT deep research: who is it best for?

ChatGPT deep research is best suited to analysts who need a downloadable, cited report from web and connected material, rather than paper discovery.

  • Company researchers can use connected sources when a project calls for approved internal material alongside web information.1
  • Academic literature reviewers should not choose ChatGPT deep research solely to find papers; a cited web report does not replace screening and reading them.

Its main advantage is an editable report workflow that connects to sources, not a guarantee of correct citations. In the same 100-task benchmark, OpenAI Deep Research scored 77.96% for citation accuracy.12

ChatGPT review Independent, data-backed
3

Gemini Deep Research

4.6/5

At a glance

  • Google says Gemini Deep Research uses the Gemini 3 model to plan and produce reports.
  • Gemini lists a pricing option at $99.99 per month.
  • Its pricing information also lists 20 TB of storage across Gmail, Drive, and Photos.

Gemini Deep Research suits company research that pairs outside information with Google Workspace notes. ChatGPT deep research has a stated workflow for prioritizing connected sources. Compare access, citation support, and the research allowance attached to a paid plan.

Gemini Deep Research: features confirmed by the vendor

Google says Gemini Deep Research uses the Gemini 3 model to plan and produce reports.15 Its sales-lead example examines a company’s products, funding, team, and competitors, then combines that work with Workspace client notes.15 That is a company-analysis workflow; on its own, it says nothing about academic-paper coverage.16

Gemini Deep Research: pricing, limits and citation quality

Gemini lists a pricing option at $99.99 per month.17 Its pricing information also lists 20 TB of storage across Gmail, Drive, and Photos. Storage is not a Deep Research report allowance.17 The reported 5x and 20x higher rate limits concern an agent model in Google Antigravity, not a stated Deep Research quota.17 Each report citation still needs checking against the claim it supports.18

Gemini Deep Research: who is it best for?

Gemini Deep Research is best suited to company analysts who work with client notes in Google Workspace. It can draw on three Workspace source types: Gmail, Drive, and Chat.19

  • Company analysts can consider Gemini Deep Research for questions that pair a prospective client’s history with existing Workspace notes.15
  • Academic literature reviewers should not infer paper-discovery coverage from a sales-lead workflow; they must assess discovery and source verification separately.

Workspace-connected company analysis is Gemini Deep Research’s clearest fit. Google AI Pro includes 5 TB of storage and expanded Deep Research access, but Google lists no fixed report count with them.18

Independent, data-backed
4

Gemini Notebook (NotebookLM)

4.6/5

At a glance

  • A free Gemini Notebook notebook holds up to 50 user-selected sources.
  • Google says NotebookLM became Gemini Notebook in July 2026; existing notebooks remain accessible.
  • An uploaded source can contain up to 500,000 words.

Gemini Notebook (NotebookLM) is better suited to selected documents than a broad web search. Researchers choose the material, then check answers against it. A free Gemini Notebook notebook holds up to 50 user-selected sources.20

Gemini Notebook (NotebookLM): features confirmed by the vendor

Gemini Notebook accepts help-center articles, product documentation, and release notes as sources.21 It can turn uploaded blog posts, reports, and webinars into infographics or Video Overviews.21 Google says NotebookLM became Gemini Notebook in July 2026; existing notebooks remain accessible.22 An uploaded source can contain up to 500,000 words.20

Gemini Notebook (NotebookLM): pricing, limits and citation quality

Gemini Notebook points users to its Help Center for usage limits; the cited vendor information does not tie a notebook or source allowance to a priced plan.21 A document citation identifies where to look, but researchers must confirm that the passage supports the claim. Gemini Deep Research storage capacity is not a Gemini Notebook research limit.

Gemini Notebook (NotebookLM): who is it best for?

Gemini Notebook is best suited to researchers and teams with a defined set of documents to analyze.

  • Documentation teams can work from help articles, product documents, and release notes in Gemini Notebook.21
  • Literature reviewers seeking new papers should not rely on a notebook of selected sources for paper discovery.

Its strength is close work with supplied material, not broad discovery.

Independent, data-backed
5

Claude Research

4.4/5

At a glance

  • Another combines web, advertising, and channel data from Amplitude and BigQuery in a review deck, flagging changes above 10%.
  • Claude’s everyday-work paid plan costs $17 per month with an annual subscription discount, billed $200 up front, or $20 per month billed monthly.

Claude Research suits analysts working across company documents and connected business data. It is less directly suited to discovering academic papers. Its value rests on accessible sources, plan cost, and a clear trail from claims to evidence.

Claude Research: features confirmed by the vendor

Claude’s vendor examples bring customer feedback from Slack, email, and support records together for review.23 Another combines web, advertising, and channel data from Amplitude and BigQuery in a review deck, flagging changes above 10%.23 These are examples, not a promise that every team can connect those sources.

Claude Research: pricing, limits and citation quality

Claude’s everyday-work paid plan costs $17 per month with an annual subscription discount, billed $200 up front, or $20 per month billed monthly.24 Those prices do not state a Claude Research report allowance. Before acting on a finding, analysts should trace it to the underlying document or data record.

Claude Research: who is it best for?

Claude Research is best suited to teams turning connected business information into a deliverable an analyst can review.

  • Business analysts can assess Claude for work combining internal feedback with performance data.23
  • Academic literature reviewers should not infer a dedicated paper-discovery process from Claude’s business examples.

Its clearest fit is business analysis across sources, with a person checking the evidence.

Claude review Independent, data-backed
6

Elicit

3.7/5

At a glance

  • Elicit searches more than 138 million academic papers and 545,000 clinical trials.
  • The vendor says it can find up to 1,000 relevant papers and analyze up to 20,000 data points at once.
  • Elicit’s deeper-research plan costs $11 per user per month, billed $132 annually.

Elicit helps researchers find, screen, and analyze academic papers rather than general web results. The right plan depends on search size, review workflow, and cost.

Elicit: features confirmed by the vendor

Elicit searches more than 138 million academic papers and 545,000 clinical trials.6 The vendor says it can find up to 1,000 relevant papers and analyze up to 20,000 data points at once.6 Those capacities suit larger searches, but researchers still need to read each study’s methods and findings.

Elicit: pricing, limits and citation quality

Elicit’s deeper-research plan costs $11 per user per month, billed $132 annually. Its systematic-reviews plan costs $39 per user per month, billed $468 annually.25 The Systematic Review Workflow can screen 5,000 papers.25 Inclusion in a review does not make a paper evidence for every summary claim. Check the relevant passage and study design.

Elicit: who is it best for?

Elicit is best suited to researchers screening and analyzing academic literature at scale.

  • Systematic-review teams should compare Elicit’s workflow with their screening volume and extraction fields.
  • Company analysts should not make Elicit their primary general-web or internal-business tool; its stated search centers on papers and clinical trials.6

Elicit offers a paper-focused workflow, but annual billing and review capacity deserve attention before purchase.

Independent, data-backed
7

Consensus

At a glance

  • Consensus searches and analyzes peer-reviewed literature from 400 million scholarly works, including licensed publisher full text.
  • Students and faculty with a valid school email can get up to 40% off a subscription.

Consensus suits questions grounded in scholarly literature. Elicit describes a more specific paper-screening workflow. Compare the works each searches, the link between answers and papers, and included use before extra charges apply.

Consensus: features confirmed by the vendor

Consensus searches and analyzes peer-reviewed literature from 400 million scholarly works, including licensed publisher full text.26 That collection spans academic fields. Size alone, though, cannot show whether an answer fairly represents the studies it cites.27

Consensus: pricing, limits and citation quality

Students and faculty with a valid school email can get up to 40% off a subscription.4 Before carrying an answer into a review, read the cited paper’s methods, population, and findings. The allowance alone does not settle a larger project’s total cost.

Consensus: who is it best for?

Consensus is best suited to students and analysts investigating scholarly questions who will read the papers behind its answers.

  • Academic researchers can search Consensus for studies relevant to a question.26
  • Systematic-review teams should not mistake a Deep Review allowance for a documented screening and extraction workflow.
  • Company researchers should look elsewhere as a primary tool if their evidence mainly comes from current web pages or internal files.

Consensus centers on scholarly search. Its answers still need paper-level checks.

Independent, data-backed
8

Scite

3.1/5

At a glance

  • Scite reports more than 323 million indexed articles.
  • Scite says agreements with 44 publisher partners contribute to its licensed scholarly collection alongside open-access content.
  • New Scite users can get a 7-day free trial of premium features.

Scite suits researchers following an idea through scholarly and related records. Consensus focuses on answers drawn from scholarly literature. Compare their coverage, citation evidence, and plan access.

Scite: features confirmed by the vendor

Scite reports more than 323 million indexed articles.28 It also describes access to patents, clinical trials, grants, preprints, and datasets for work extending beyond journal papers.28 Scite says agreements with 44 publisher partners contribute to its licensed scholarly collection alongside open-access content.28,29

Scite: pricing, limits and citation quality

New Scite users can get a 7-day free trial of premium features.30 The deep-research and API tier includes a $60 API credit allowance, not a stated report count.30 A citation alone does not validate a result. Read the passage to see whether it supports the particular claim.

Scite: who is it best for?

Scite is best suited to researchers following scholarly work across publications and related records.

  • Academic researchers can assess Scite for searches involving preprints, patents, or datasets as well as papers.28
  • Company analysts seeking current web coverage should not confuse Scite’s range of research records with general web search.

Cross-record coverage is Scite’s strength, and the trial offers a chance to assess its fit before paying.

Independent, data-backed
9

SciSpace

At a glance

  • SciSpace shares a customer testimonial about its appeal to someone without specialist scientific training.
  • Compare SciSpace with other paper tools using the literature your work calls for.

SciSpace is an academic research option. Judge it by the papers you need, the evidence behind its answers, and paid-plan terms. A customer testimonial about accessible scientific information expresses an opinion, not a verified accuracy measure.7

SciSpace: features confirmed by the vendor

SciSpace shares a customer testimonial about its appeal to someone without specialist scientific training.7 That account establishes neither a searchable-paper count nor citation-checking or review capacity. Compare SciSpace with other paper tools using the literature your work calls for.

SciSpace: pricing, limits and citation quality

SciSpace says users can cancel anytime.7 The refund condition covered above concerns a purchase, not a research-use allowance. Before relying on a SciSpace summary, check its citations against the original papers.

SciSpace: who is it best for?

SciSpace suits readers exploring scientific literature who are willing to inspect the papers themselves.

  • Readers new to scientific literature can assess whether SciSpace helps explain a paper without treating a testimonial as accuracy evidence.
  • Systematic-review teams should base their decision on documented screening and citation needs, not cancellation terms or a testimonial.

Flexible purchase terms cannot stand in for verified research output.

Independent, data-backed
10

Semantic Scholar

At a glance

  • Semantic Scholar suits scientific-paper discovery.
  • Semantic Scholar indexes more than 200 million academic papers through publisher partnerships, data providers, and web crawls.

Semantic Scholar suits scientific-paper discovery. Scite describes coverage extending to patents and datasets. Compare source scope, access to underlying papers, and task-specific limits.

Semantic Scholar: features confirmed by the vendor

Semantic Scholar indexes more than 200 million academic papers through publisher partnerships, data providers, and web crawls.31 Semantic Reader works with most arXiv papers and lets readers highlight text for annotations.32 These tools support discovery and close reading, not research into current company web pages.33

Semantic Scholar: pricing, limits and citation quality

The size of Semantic Scholar’s index measures discovery coverage, not a report allowance or price. Its nonprofit research-institute background is not a pricing term either.31 To judge a citation, open the original paper and compare its methods and findings with the claim.

Semantic Scholar: who is it best for?

Semantic Scholar is best suited to researchers finding scientific papers and reading relevant arXiv work closely.

  • Students and academics can locate papers with Semantic Scholar and annotate supported arXiv papers while reading.32
  • Company analysts should not make an academic-paper index their main source for current web or internal-business research.31

Semantic Scholar helps people find and read papers; it does not remove the need to check evidence.

Independent, data-backed
11

ResearchRabbit

At a glance

  • ResearchRabbit supports literature discovery through collections and collaboration.
  • ResearchRabbit’s Free Forever tier includes unlimited search and collections, collaboration, and library uploads.

ResearchRabbit supports literature discovery through collections and collaboration. Semantic Scholar puts more emphasis on searching and reading papers. Workflow, free-tier limits, and verification needs will decide the better fit.

ResearchRabbit: features confirmed by the vendor

ResearchRabbit’s Free Forever tier includes unlimited search and collections, collaboration, and library uploads.34 On that tier, researchers can start discovery with up to 50 seed articles.34 Collections help organize a review, but a related paper still needs to support the claim for which it is cited.

ResearchRabbit: pricing, limits and citation quality

ResearchRabbit+ is a monthly subscription at the US price in the comparison table, not an annual-billing offer.34 Subscribers can cancel and keep paid access until the plan ends.34 For a large starting collection, the free tier’s seed-article cap matters. Every paper still needs a findings check before citation.

ResearchRabbit: who is it best for?

ResearchRabbit is best suited to researchers organizing and expanding a shared literature search before selecting usable evidence.

  • Literature-review teams can coordinate paper discovery with ResearchRabbit’s collections and collaboration features.34
  • Company analysts should not use a scholarly-literature workflow as their primary tool for current web or internal-business research.

ResearchRabbit helps discover papers, not verify what they prove.

Independent, data-backed
12

Undermind

3.8/5

At a glance

  • The vendor used 23 complex research queries; that result cannot guarantee performance on every topic.
  • An example result is a Zhao et al. paper on resistance to KRAS(G12C) inhibition.
  • Undermind’s evaluation compares recall, the share of relevant papers found, after 5 and 10 minutes of searching.

Undermind suits narrowly defined questions that call for relevant-paper retrieval. ResearchRabbit focuses on expanding a search from seed articles. Compare retrieval, evidence behind results, and project usage terms.

Undermind: features confirmed by the vendor

Undermind claims its evaluation found twice as many relevant papers as AI web search.35 The vendor used 23 complex research queries; that result cannot guarantee performance on every topic.35 An example result is a Zhao et al. paper on resistance to KRAS(G12C) inhibition. Read the study before citing it.35

Undermind: pricing, limits and citation quality

Undermind’s evaluation compares recall, the share of relevant papers found, after 5 and 10 minutes of searching. It also considers precision, the share of retrieved papers that are relevant.35 Those intervals are benchmark measures, not subscription limits or report allowances. A ranked result does not prove that a paper supports a particular claim.

Undermind: who is it best for?

Undermind is best suited to researchers seeking relevant papers for complex questions.

  • Specialist literature researchers can assess Undermind with a defined question, then read the papers it retrieves.
  • Company analysts seeking current web information should not take a paper-retrieval benchmark as evidence of general web coverage.

Targeted retrieval is Undermind’s stated strength. Its benchmark cannot replace reading the papers.

Independent, data-backed

Methodology and limits of this comparison

AI research tool rankings bring web-research and academic-literature products together by source scope, traceable citations, plan cost, and usage limits. Positions follow the supplied product order. They are neither scores nor results from a head-to-head test.

This comparison draws on public vendor pages, pricing pages, and documentation available as of publication, without hands-on testing. Vendor benchmarks and testimonials remain vendor claims, not independently verified findings. A citation feature helps identify a source; it cannot make every generated claim correct.

The practical verdict starts with the task, not the ranking position: web reports, selected documents, and academic-paper discovery call for different sources.

Web research vs academic literature tools

Web research tools suit company and current-information questions. Academic literature tools suit questions requiring studies to be found and evaluated. They differ in source coverage, workflow, and output. A source-grounded notebook handles another job: examining material a researcher has chosen.

The table pairs question types with research workflows and tools from this ranking.

Question type Tool type Example on this list
What is known about a company? Web research Gemini Deep Research
What do these selected documents say? Source-grounded notebook Gemini Notebook (NotebookLM)
Which studies address a research question? Academic literature search Elicit, Consensus, ResearchRabbit
What does a paper say beyond its abstract? Full-text research and reading Scite, SciSpace

Gemini Deep Research can combine web research with a user’s Gmail, Drive, Chat, and uploaded files when the user provides access.15 Elicit uses semantic search to find related papers and clinical trials without exact keyword matches.6 ResearchRabbit offers access to more than 320 million academic papers for discovery.36,37

Output workflow matters as much as collection size. Gemini Deep Research can turn reports into Audio Overviews or interactive Canvas content, while Scite searches beyond a paper’s abstract into full text.15,28 SciSpace offers PDF chat and an AI reading feature that explains highlighted passages in simpler terms.7 Clear presentation still leaves readers responsible for checking each claim against its source.

The choice is web research for current external questions, a notebook for selected files, or literature tools for finding and reading papers.

How to verify AI research output

AI research output is ready to use only when someone has checked consequential claims against original sources and recorded the evidence. Citations help with that work. They do not do it for the researcher.

  • Open every citation. Perplexity uses inline citations, and Elicit supplies sentence-level citations. Compare the cited passage with the precise claim it supports.6,10
  • Check a paper’s status and context. Look for retractions, then read methods and findings before citing a conclusion. Scite’s Smart Citations identify later papers that support, challenge, or discuss a work; its index has more than 1.6 billion scholarly citations.28,29,30
  • Compare findings across tools. For questions spanning current information and published studies, use web and literature tools, then resolve disagreements through primary sources.
  • Keep a source log. Record each claim, source title, supporting passage, publication details, and unresolved conflict so another reader can retrace the evidence.

Verified research connects claims to supporting evidence, not just plausible citations; AI-generated output is not medical or legal advice.

What to check before you pay

AI research subscriptions deserve a close look at permitted work, total billing commitment, and rights over sources and output. ChatGPT Free has limited Deep Research access, Plus expands it, and Pro has the highest listed level.13 Consensus Free offers basic paper search and limited AI analysis; paid plans raise Deep Review allowances.4

  • Free-tier access should let you evaluate your intended task. Confirm that the trial or free plan supports the report, paper search, or document analysis you need.
  • Usage caps and credits should match expected work. Undermind reports an average production deep-search duration of 2.5 minutes, but duration is not a monthly allowance.35
  • Commercial-use rights should cover how your team shares or publishes results. Review terms for generated text, uploaded documents, and third-party material.
  • Data-use terms should fit your organization’s rules for confidential files. Check whether a provider may use uploads for model training before connecting private sources.
  • Billing and cancellation terms should be judged by total commitment, not just a monthly equivalent. Include expected usage charges and the required billing period.

The purchase decision should follow a matched free-tier task and a review of usage, rights, data handling, and total cost.

Sources

  1. ↑Głębokie badanie w ChatGPT https://chatgpt.com/features/deep-research/
  2. ↑OpenAI, Introducing deep research https://openai.com/index/introducing-deep-research/
  3. ↑OpenAI Help Center, What is ChatGPT Plus? https://help.openai.com/en/articles/6950777-what-is-chatgpt-plus
  4. ↑Pricing https://consensus.app/pricing/
  5. ↑Consensus Help Center, Subscription Plans https://help.consensus.app/en/articles/10087865-subscription-plans
  6. ↑Elicit: AI for research & decision-making https://elicit.com/
  7. ↑SciSpace Premium – Unlimited access to AI research tools https://scispace.com/pricing
  8. ↑Elicit Systematic Reviews help center and SciSpace refund policy. ([support.elicit.com](support.elicit.com https://scispace.com/help/en/articles/14686552-how-to-request-a-refund-on-scispace
  9. ↑Perplexity, Pro Perks https://www.perplexity.ai/properks
  10. ↑Cennik Perplexity: plany dla użytkowników indywidualnych i Enterprise https://www.perplexity.ai/hub/pricing
  11. ↑Perplexity, Research with premium sources https://www.perplexity.ai/premium-sources
  12. ↑DeepResearch Bench https://deepresearch-bench.github.io/
  13. ↑Plany i ceny | ChatGPT https://chatgpt.com/pricing/
  14. ↑OpenAI’s deep research announcement, current deep research help page, and ChatGPT release notes. ([openai.com](openai.com https://help.openai.com/en/articles/10500283-research-faq
  15. ↑Gemini Deep Research — your personal research assistant https://gemini.google/us/overview/deep-research/?hl=en
  16. ↑Google Gemini Deep Research product page. ([gemini.google](gemini.google https://gemini.google/overview/deep-research/
  17. ↑Google AI Pro & Ultra — get access to Gemini 4 Argon & more https://gemini.google/us/subscriptions/?hl=en
  18. ↑Google One, Google AI plans. ([one.google.com](one.google.com https://one.google.com/intl/en/about/google-ai-plans/?gl=US&hl=en
  19. ↑Google, Google Workspace apps can now be integrated into Deep Research https://blog.google/products-and-platforms/products/gemini/deep-research-workspace-app-integration/
  20. ↑Google, Add or discover new sources for your notebook https://support.google.com/gemininotebook/answer/16215270?hl=en
  21. ↑Gemini Notebook | Premium AI Research & Brainstorming Tool https://notebook.google/plans?hl=en
  22. ↑Gemini Notebook | AI Research Tool & Thinking Partner https://notebook.google/?hl=en
  23. ↑Claude | The AI assistant for work https://claude.com/product/overview
  24. ↑Plans & pricing | Claude by Anthropic https://claude.com/pricing
  25. ↑Pricing | Elicit: The AI Research Assistant https://elicit.com/pricing
  26. ↑Consensus: AI for Research https://consensus.app/
  27. ↑Consensus, Research corpus. ([consensus.app](consensus.app https://consensus.app/home/features/corpus/
  28. ↑AI for Research | Scite https://scite.ai/
  29. ↑Scite coverage and data pages. ([scite.ai](scite.ai https://scite.ai/coverage
  30. ↑Scite Pricing https://scite.ai/pricing
  31. ↑Semantic Scholar | About Us https://www.semanticscholar.org/about
  32. ↑Semantic Scholar | Semantic Reader https://www.semanticscholar.org/product/semantic-reader
  33. ↑Semantic Scholar librarian resources and Semantic Reader product page. ([webflow.semanticscholar.org](webflow.semanticscholar.org https://webflow.semanticscholar.org/product/semantic-reader
  34. ↑Pricing | ResearchRabbit https://www.researchrabbit.ai/pricing
  35. ↑Undermind: your AI co-researcher for the literature https://www.undermind.ai/
  36. ↑ResearchRabbit: AI Tool for Smarter, Faster Literature Reviews https://www.researchrabbit.ai/
  37. ↑Google Workspace Deep Research help, Elicit API documentation, and ResearchRabbit product page. ([support.google.com](support.google.com https://support.google.com/mail/answer/17127708?hl=en

Prices and terms checked at the source: November 2026.