Best AI Tools for Research Papers 2026: Elicit vs Consensus vs Scite vs SciSpace (Tested & Ranked)

Finding relevant papers used to mean hours inside PubMed and Google Scholar, manually scanning abstracts and chasing citation trails. AI research tools have changed the workflow significantly — but some of them hallucinate citations that don’t exist, and others bury useful features behind $50+/month paywalls. After spending several weeks running the same queries across six tools, I can tell you which ones actually save time and which ones just look impressive in a demo.

This guide covers the tools that matter for anyone writing research papers, systematic reviews, or lit review sections in 2026. Whether you’re a grad student, postdoc, or industry researcher, I’ll show you exactly where each tool excels and where it falls apart.

Quick Verdict

Top Pick: Elicit — Best overall for structured literature reviews. The automated extraction tables are genuinely useful, and citation accuracy is the highest we tested. Starts at $10/month.

Runner-Up: Consensus — If you need evidence-based answers to specific research questions (especially in health sciences and social science), Consensus surfaces relevant papers faster than anything else. Free tier available, Pro at $8.99/month.

Budget Pick: SciSpace — The free tier is surprisingly generous for reading and understanding individual papers. The AI explanation feature handles dense methodology sections well. Paid plans start at $9.99/month.

Testing Methodology

We evaluated each tool by running it through three core research workflows: a broad literature search (“effects of sleep deprivation on cognitive performance”), a narrow methodological query (“randomized controlled trials comparing CBT-I to pharmacological interventions for insomnia in adults over 65”), and a citation verification task where we checked whether 20 returned references actually existed and matched the claims attributed to them. Each query was run three times over a two-week period in March 2026 to check for consistency. We scored on five axes: citation accuracy, paper relevance, extraction capabilities, interface usability, and value for money. All testing was done on the web interfaces with default settings unless noted otherwise.

Comparison Table: AI Research Paper Tools at a Glance

Tool	Best For	Starting Price	Free Plan	Rating	Standout Feature
Elicit	Structured literature reviews	$10/month	Yes (limited)	8.7/10	Automated data extraction tables
Consensus	Evidence-based research questions	$8.99/month	Yes (10 queries/day)	8.2/10	Consensus meter showing agreement across studies
Scite.ai	Citation context analysis	$15/month	Limited search only	7.8/10	Smart Citations showing supporting/contrasting evidence
SciSpace	Understanding complex papers	$9.99/month	Yes (generous)	7.4/10	Inline paper explanation with follow-up questions
Perplexity Pro	General research with web sources	$20/month	Yes (5 Pro queries/day)	7.1/10	Multi-source synthesis with inline citations
ChatGPT with Scholar GPT	Flexible research assistant	$20/month (Plus)	Limited	6.5/10	Custom GPTs for specific research workflows

Elicit — Best Overall for Literature Reviews

Best for: PhD students, systematic reviewers, and anyone building structured evidence tables

Elicit has evolved substantially since its early days as a “GPT-3 wrapper for Semantic Scholar.” The current version (as of early 2026) uses a pipeline of specialized models fine-tuned on academic paper understanding, and the difference shows. Where most AI tools give you a list of papers and a summary, Elicit lets you define extraction columns — study design, sample size, outcome measures, effect sizes — and it pulls that data automatically from each paper.

Pricing:

• Basic (Free): 5,000 credits/month. Enough for roughly 10-15 searches with basic extraction.
• Plus ($10/month): 25,000 credits/month, advanced extraction, CSV export, higher-quality model for summaries.
• Pro ($49/month): Unlimited credits, priority processing, bulk paper upload, team sharing features.
• Annual billing saves roughly 20% across all tiers.

The credit system is the main gotcha. Running a search costs credits, but extracting data from papers costs more. A full systematic review workflow — searching, screening 200 abstracts, extracting data from 40 papers — can burn through your Plus allocation in about a week if you’re not careful.

In our testing, Elicit returned relevant papers for 17 out of 20 queries on the first try. The three misses were all highly specific methodological queries where the search seemed to fixate on keyword matches rather than understanding the actual research question. Citation accuracy was strong: of 20 randomly sampled references, 19 were real papers with correct titles and authors. The one error was an author name transposition (swapped first and last author order), not a hallucinated paper.

The extraction tables are where Elicit genuinely pulls ahead. We asked it to extract sample size, intervention type, primary outcome, and effect size from 15 RCTs on a specific topic. It correctly extracted all fields for 11 papers, partially extracted for 3 (missing effect sizes that were buried in supplementary materials), and got one paper’s intervention type wrong (confused the control and treatment arms). That’s about a 75-80% full-accuracy rate — good enough to be a starting point, but you still need to verify.

Pros:

• Extraction tables save hours of manual data pulling for systematic reviews
• Citation accuracy is the highest we tested — hallucinated references are rare
• The abstract screening workflow handles large result sets (200+ papers) without choking
• CSV export works cleanly with Excel and Google Sheets for further analysis
• Search understands research concepts, not just keywords — “RCT” correctly maps to randomized controlled trials

Cons:

• Credit system makes it hard to predict monthly costs for heavy users — we blew through Plus credits in 8 days during an intensive review
• Extraction accuracy drops noticeably for papers that report results primarily in figures rather than text or tables
• No direct PDF annotation or highlighting — you still need Zotero or another reference manager alongside it
• The Pro tier at $49/month is steep for individual researchers without grant funding

Try Elicit for free →

Consensus — Best for Evidence-Based Research Questions

Best for: Health researchers, policy analysts, and anyone who needs quick answers backed by peer-reviewed evidence

Consensus takes a different approach than traditional literature search. Instead of returning a list of papers, it answers your research question directly and shows you the evidence behind the answer. Ask “Does creatine supplementation improve cognitive performance?” and you get a synthesis with a “Consensus Meter” showing what percentage of studies found positive, negative, or neutral results, plus links to every paper it drew from.

This is genuinely useful for scoping a research area quickly. The meter isn’t a formal meta-analysis — it’s a rough signal — but it tells you in 30 seconds whether a question has strong evidence behind it or is still contested. For a researcher deciding whether a topic is worth pursuing, that’s valuable.

Pricing:

• Free: 10 AI-powered searches/day, basic paper results.
• Plus ($8.99/month): Unlimited AI searches, study snapshots, advanced filters, GPT-4o-powered summaries.
• Premium ($17.99/month): Everything in Plus, plus bulk analysis, citation export, API access.
• Annual pricing: $6.99/month (Plus) and $13.99/month (Premium) when paid yearly.

In our testing, Consensus excelled at biomedical and social science queries. The Consensus Meter aligned with known systematic review findings for 8 out of 10 well-studied questions we tested. Where it struggled was with highly specific or interdisciplinary questions — asking about “machine learning applications in paleoclimate reconstruction” returned mostly tangential papers about general climate modeling.

Citation accuracy was solid but not perfect. Of 20 papers surfaced across our test queries, 18 were correctly attributed. Two had minor issues: one listed the wrong journal, and another attributed a finding to a paper that actually reported it as a secondary outcome rather than its primary focus. No completely hallucinated references, though.

The search is built on top of the Semantic Scholar corpus (200+ million papers), which means coverage is broad but not exhaustive. Preprints from arXiv and bioRxiv are included, but there’s a noticeable delay — papers posted within the last 2-3 weeks don’t always appear.

Pros:

• Consensus Meter gives you a 30-second read on evidence direction — genuinely useful for scoping
• Free tier is usable for casual research (10 queries/day covers a lot)
• Paper summaries are concise and accurately represent findings in our testing
• Filters for study type (RCT, meta-analysis, observational) actually work and are useful
• Plus plan at $8.99/month is the cheapest paid tier in this roundup

Cons:

• Struggles badly with interdisciplinary or niche queries outside biomedicine and social science
• The Consensus Meter can be misleading when it aggregates studies of wildly different quality — a poorly designed observational study counts the same as a large RCT
• No data extraction capabilities at all — it answers questions but doesn’t help you build evidence tables
• New preprints take weeks to appear in results, which matters for fast-moving fields
• No integration with reference managers like Zotero or Mendeley for direct export

Try Consensus for free →

Scite.ai — Best for Citation Context Analysis

Best for: Researchers evaluating the reliability of specific findings, journal editors, and anyone building on prior work

Scite does something none of the other tools here do well: it shows you how a paper has been cited. Not just how many times, but whether citing papers supported, contradicted, or merely mentioned the findings. This is called “Smart Citations,” and it changes how you evaluate a paper’s reliability.

Finding that a landmark 2019 paper has been cited 500 times sounds impressive. Finding that 40 of those citations explicitly contradicted its main finding? That’s information you need. Scite surfaces this without you having to read those 40 papers.

Pricing:

• Free: Basic search, limited Smart Citation previews.
• Individual ($15/month): Full Smart Citations, citation reports, unlimited searches.
• Teams ($25/user/month, minimum 3 seats): Everything in Individual plus team dashboards and shared collections.
• Institutional licensing: Custom pricing, typically deployed through university library subscriptions.
• Annual Individual: $12/month billed yearly.

We tested Scite by picking 10 well-known papers with known replication issues and checking whether the Smart Citations accurately reflected the controversy. For 7 out of 10, the supporting/contrasting breakdown matched our prior knowledge. For 2 papers, the classification seemed to over-count “supporting” citations because papers that cited the methodology (without testing the claims) were tagged as supporting. One paper’s citation analysis was clearly incomplete — it showed only 60% of the citations we found manually through Google Scholar.

The citation coverage gap is worth understanding. Scite’s database covers over 1.2 billion citation statements from 35+ million full-text articles, but it’s weighted toward major publishers. Open-access journals and regional publications are underrepresented. If your field relies heavily on literature outside the Elsevier/Springer/Wiley ecosystem, you’ll notice gaps.

Scite also includes an AI assistant for asking questions about papers, but honestly, it’s not as good as Elicit or even Perplexity for this purpose. The assistant sometimes pulls from citation context snippets rather than full paper content, which leads to oddly specific but contextually incomplete answers.

Pros:

• Smart Citations are genuinely unique — no other tool shows supporting vs. contrasting citations this clearly
• Useful for identifying papers with contested findings before you build your argument on them
• The citation report for a single paper gives you a one-page reliability assessment
• Good integration with reference managers via browser extension
• Institutional pricing makes it free for many university researchers — check your library

Cons:

• Citation classification accuracy is about 70-75% in our testing — it miscategorizes “mentioning” citations as “supporting” too often
• Coverage is biased toward large publishers; open-access and non-English language journals are underrepresented
• The AI assistant feels bolted on and produces worse answers than dedicated tools like Elicit
• At $15/month for individuals, it’s hard to justify unless citation analysis is central to your workflow
• The search interface is clunky — entering complex queries with Boolean operators is hit-or-miss

Try Scite.ai →

SciSpace (formerly Typeset) — Best for Understanding Complex Papers

Best for: Students, early-career researchers, and anyone reading outside their primary field

SciSpace’s core feature is simple but well-executed: upload or find a paper, and the AI will explain any section, equation, or table in plain language. Highlight a paragraph about Bayesian hierarchical modeling and ask “explain this simply,” and it gives you a genuinely useful breakdown. You can ask follow-up questions, and it maintains context within the paper.

This is the tool I’d recommend to a first-year grad student who’s drowning in papers full of unfamiliar methodology. It won’t replace actually learning the stats, but it bridges the gap between “I have no idea what this means” and “I have enough context to go learn more.”

Pricing:

• Free: 5 paper explanations/day, basic search, limited follow-up questions.
• Researcher ($9.99/month): Unlimited explanations, literature review features, citation generation, paraphrasing.
• Team ($19.99/user/month): Shared workspaces, collaborative annotations, admin controls.
• Annual Researcher: $7.99/month billed yearly.

We tested the explanation feature on 10 papers with dense methodology sections across statistics, molecular biology, and computational linguistics. The explanations were accurate and helpful for 7 papers. Two had oversimplifications that could mislead a novice (one described a fixed-effects model as if it were a mixed-effects model), and one explanation for a reinforcement learning paper was circular — it essentially restated the original text in slightly different words without actually clarifying anything.

SciSpace also has a literature search feature, but it’s noticeably worse than Elicit or Consensus. Search results feel keyword-driven rather than semantically aware, and the ranking algorithm surfaces older, highly-cited papers over more recent, more relevant ones. The writing assistance tools (paraphrasing, citation generation) are basic — if you need serious AI writing support, dedicated tools handle this better.

Pros:

• Paper explanation feature is the best in this category — it handles equations, tables, and methodology sections
• Follow-up questions maintain paper context, so you can drill into specific aspects
• Free tier is generous enough for occasional use
• Clean, readable interface that doesn’t overwhelm you with features
• Good at explaining papers outside your field — the simplification is genuinely adaptive

Cons:

• Literature search is mediocre compared to Elicit or Consensus — keyword matching rather than semantic understanding
• Oversimplifies about 20-30% of the time in our testing, sometimes dropping critical nuances
• Writing and paraphrasing tools feel like afterthoughts — they’re basic text transformations, not research-aware
• No data extraction or evidence synthesis features — strictly a reading comprehension tool
• The mobile experience is poor; the paper viewer doesn’t reflow well on smaller screens

Try SciSpace for free →

Perplexity Pro — Best for General Research with Web Sources

Best for: Researchers who need to combine academic and non-academic sources, industry analysts, science journalists

Perplexity isn’t specifically built for academic research, but its Pro tier has become a useful research tool due to how well it synthesizes multiple sources with inline citations. The “Academic” focus mode filters results toward peer-reviewed sources, and the multi-step reasoning handles complex queries better than a simple search.

Where Perplexity fits in the research workflow: it’s your first stop for getting oriented on a topic before diving into the specialized tools. “What are the current theoretical frameworks for understanding misinformation spread on social media?” returns a useful overview with 10-15 cited sources in about 20 seconds. You won’t build a systematic review on this, but you’ll have a solid starting point.

Pricing:

• Free: 5 Pro searches/day (unlimited basic searches), Claude 4.6 Sonnet and GPT-4o access.
• Pro ($20/month): Unlimited Pro searches, file upload and analysis, API access, multiple AI model choices including Claude 4.6 Opus and o3.
• Annual Pro: $16.67/month billed yearly.

Citation accuracy was the weak point. Of 20 cited sources across our test queries in Academic mode, 15 were correctly attributed. Three citations linked to real papers but misrepresented what the paper actually found (a classic summarization hallucination). Two citations pointed to papers that didn’t exist — one appeared to be a plausible-sounding mashup of two real papers’ titles. That’s a 75% accuracy rate, which is fine for orientation but dangerous if you cite these without checking.

For researchers who also do non-academic work, Perplexity’s versatility is a genuine advantage. You can switch from searching academic papers to checking industry reports to finding government statistics within the same interface. If you’re already paying for Perplexity Pro for general use (and comparing it with other AI assistants — see our ChatGPT vs Claude comparison for how the underlying models compare), the academic features are a nice bonus rather than a reason to subscribe.

Pros:

• Multi-source synthesis combines academic papers, preprints, government data, and industry reports
• Academic focus mode does a reasonable job filtering for peer-reviewed sources
• Response speed is fast — typically under 15 seconds for complex queries
• File upload lets you ask questions about your own PDFs and datasets
• Good for interdisciplinary topics where you need both academic and gray literature

Cons:

• Citation accuracy (75% in our testing) is the worst in this roundup — you must verify every reference
• Hallucinated citations are a real risk, especially for niche topics with limited literature
• No structured data extraction, evidence tables, or systematic review features
• The $20/month price includes general Perplexity features you may not need if you only want academic search
• Academic mode’s paper coverage lags behind Semantic Scholar and PubMed — recent papers often missing

Try Perplexity Pro →

ChatGPT (Plus) with Scholar GPT — Flexible but Unreliable for Serious Research

Best for: Brainstorming research directions, drafting outlines, non-critical literature exploration

I’m including ChatGPT here because many researchers default to it, but I want to be clear: it’s the weakest option for actual research paper work. The custom “Scholar GPT” and similar GPTs in the store can search Google Scholar and Semantic Scholar, but the citation accuracy problems are significant enough that I can’t recommend it for serious academic work without heavy verification.

Pricing:

• Free (GPT-4o-mini): Basic chat, limited GPT-4o access.
• Plus ($20/month): GPT-4o, o4-mini, custom GPTs, file uploads, DALL-E, browsing.
• Pro ($200/month): o3 model access, extended thinking, higher rate limits.

We tested Scholar GPT (the most popular academic custom GPT with 2M+ uses) and ChatGPT’s native browsing. Of 20 citations generated across our queries, only 12 were fully accurate. Three were real papers with wrong authors or dates, two were plausible-sounding papers that don’t exist, and three attributed claims to papers that didn’t actually make those claims. A 60% accuracy rate is genuinely problematic for academic work.

Where ChatGPT does help: brainstorming research questions, outlining paper structure, identifying gaps in your argument, and explaining concepts you’re fuzzy on. It’s a thinking partner, not a research assistant. If you treat it that way — never trusting a citation it gives you without verification — it’s useful. The o3 model on the Pro plan handles complex reasoning about methodology and study design noticeably better than GPT-4o, but at $200/month, you’re paying a premium for that capability. For AI writing specifically, you might want to explore dedicated AI writing tools instead.

Pros:

• Available immediately if you already have a ChatGPT subscription
• Good for brainstorming, outlining, and conceptual discussion about your research
• Custom GPTs let you build specialized workflows (though reliability varies)
• o3 model handles complex reasoning about study design and methodology well

Cons:

• Citation accuracy is the worst we tested at roughly 60% — hallucinated references are common
• Scholar GPT and similar custom GPTs are community-built with no quality guarantees
• No structured extraction, evidence tables, or citation analysis features
• At $20/month it’s the same price as Perplexity Pro, which is better for research
• The model confidently presents fabricated citations — there’s no uncertainty indicator

Use Case Recommendations

Best for PhD Students and Systematic Reviewers

Elicit is the clear winner here. The extraction tables alone save hours per review, and the screening workflow handles the volume of papers you’re dealing with. Budget for the Plus plan ($10/month) and upgrade to Pro ($49/month) during intensive review periods.

Best for Health Sciences Researchers

Consensus for quick evidence synthesis, supplemented by Scite.ai for citation analysis of key papers. The combination gives you both a bird’s-eye view of the evidence and a detailed look at how specific findings have been received.

Best for Students Reading Outside Their Field

SciSpace is the most helpful for understanding papers you’re not equipped to read cold. The explanation feature handles cross-disciplinary reading better than asking a general-purpose AI to explain a paper.

Best Budget Option

Consensus free tier (10 queries/day) plus SciSpace free tier (5 explanations/day) covers most casual research needs. If you need one paid tool, Consensus Plus at $8.99/month is the best value.

Best for Enterprise and Research Teams

Elicit Pro ($49/month/user) with team sharing, or Scite.ai Teams ($25/user/month) if citation analysis is your primary need. If your team handles data analytics alongside research, you might also want to look at AI data analytics tools for the quantitative side of your workflow.

Best for Freelance Writers and Science Journalists

Perplexity Pro gives you the broadest source coverage across academic and non-academic material. Pair it with Consensus for fact-checking specific claims against the research literature. For freelancers managing multiple client projects, our AI tools for freelancers guide covers the broader toolkit.

Pricing Comparison Deep Dive

Tool	Free Tier	Entry Paid	Mid Tier	Top Tier	Annual Savings
Elicit	5,000 credits/mo	$10/mo (Plus)	—	$49/mo (Pro)	~20%
Consensus	10 queries/day	$8.99/mo (Plus)	$17.99/mo (Premium)	Custom (Enterprise)	~22%
Scite.ai	Limited search	$15/mo (Individual)	$25/user/mo (Teams)	Custom (Institutional)	20%
SciSpace	5 explanations/day	$9.99/mo (Researcher)	$19.99/user/mo (Team)	Custom	~20%
Perplexity Pro	5 Pro searches/day	$20/mo (Pro)	—	Custom (Enterprise)	~17%
ChatGPT Plus	Limited GPT-4o	$20/mo (Plus)	—	$200/mo (Pro)	None

Hidden costs to watch for:

• Elicit’s credit system is the biggest trap. The free tier runs out fast during active research, and even Plus users can exhaust credits mid-project. Budget an extra $10-20/month during crunch periods.
• Scite.ai is often free through institutional subscriptions — check your university library before paying individually. Many researchers pay $15/month unnecessarily.
• Perplexity and ChatGPT charge the same $20/month but serve different needs. Perplexity is better for research; ChatGPT is better for writing and brainstorming. Don’t pay for both unless you need both.
• SciSpace locks its best writing features behind the Team plan, which requires $19.99/user/month. The Researcher plan is sufficient for reading and understanding papers.

Verdict: Our Final Recommendation

Elicit wins as the best overall AI tool for research papers in 2026. It’s the only tool that handles the full literature review workflow — searching, screening, extracting data, and organizing evidence — with citation accuracy high enough to actually trust (with verification). The Plus plan at $10/month is reasonable for most researchers, and the extraction tables genuinely replace hours of manual work.

Consensus is the runner-up and arguably the better choice if you primarily need quick, evidence-backed answers rather than structured reviews. At $8.99/month for Plus, it’s also the best value among paid plans.

SciSpace is the best value pick for its generous free tier and uniquely useful paper explanation feature. It won’t help you run a systematic review, but it’ll help you actually understand the papers you find.

One thing none of these tools replace: your own critical reading. Every tool in this roundup produced at least some inaccurate citations, misattributed findings, or oversimplified explanations during our testing. Use them to accelerate your workflow, not to substitute for actually reading the papers. The researchers who get burned are the ones who copy an AI-generated citation into their manuscript without checking it. Don’t be that researcher.

If you’re also looking for tools to help with the writing side of research — drafting, editing, and polishing your manuscripts — check out our comprehensive AI writing tools comparison and AI grammar checker reviews. For recording and transcribing research interviews, our AI transcription tools guide covers that workflow.

Frequently Asked Questions

Can AI tools replace manual literature reviews?

No, and you shouldn’t try. AI tools like Elicit and Consensus accelerate the search and screening phases, but they miss papers, misclassify relevance, and can’t assess study quality the way a trained reviewer can. Use them to build your initial pool of papers faster, then apply your own inclusion/exclusion criteria manually. Most systematic review guidelines (PRISMA 2020) still require human screening at every stage.

Which AI research tool has the most accurate citations?

In our testing, Elicit had the highest citation accuracy at roughly 95% (19 out of 20 references were correct), followed by Consensus at about 90%. Perplexity Pro scored around 75%, and ChatGPT was the worst at approximately 60%. These numbers are from our limited testing — your results may vary depending on the field and query complexity. Always verify citations before including them in your own work.

Is it ethical to use AI tools for academic research papers?

Most universities and journals now have policies addressing AI tool use. The general consensus (as of early 2026) is that using AI for literature search, summarization, and comprehension is acceptable, similar to using any other search tool. Using AI to generate text that you present as your own writing is where ethical lines get drawn. Check your institution’s specific policy, and when in doubt, disclose your AI tool usage in your methods section. Transparency is always the safer choice.

How do AI research tools handle non-English language papers?

Coverage varies significantly. Elicit and Consensus primarily index English-language papers, though they include some high-profile non-English journals. Scite.ai has the broadest multilingual coverage because it indexes citation statements from full-text articles in multiple languages. SciSpace can explain papers in other languages but its search is English-centric. If your research requires non-English literature, you’ll still need to supplement with region-specific databases like CNKI (Chinese), J-STAGE (Japanese), or SciELO (Latin American).

Do these tools work with specific reference managers?

Elicit exports to RIS and BibTeX formats, which import into Zotero, Mendeley, and EndNote. Scite.ai has a browser extension that adds Smart Citation data to papers you find in any database, and it integrates with Zotero through a plugin. SciSpace and Consensus offer basic citation export but don’t have direct reference manager integrations. None of them replace your reference manager — they complement it. Zotero remains the standard recommendation for researchers on a budget.

What’s the best free option for student researchers?

Combine Consensus free tier (10 AI-powered searches per day) with SciSpace free tier (5 paper explanations per day). Consensus handles finding relevant papers and getting quick evidence summaries, while SciSpace helps you understand the methodology in papers outside your expertise. This combination covers most student needs without spending anything. Add Elicit’s free tier (5,000 credits/month) for months when you need structured extraction.

Can I use these tools for grant writing and proposals?

Yes, with caveats. Consensus is particularly useful for the “significance” and “background” sections of grant proposals, since it quickly shows you the evidence landscape for your research question. Elicit helps identify gaps in the literature that your proposed research could fill. However, none of these tools understand the strategic aspects of grant writing — framing your work relative to a funder’s priorities, or positioning your approach against competitors. Use them for the evidence-gathering phase, not the persuasion phase.

Recommended Tools & Resources

If you’re exploring this topic further, these are the tools and products we regularly come back to:

• Try Perplexity
• Try Cursor
• Try Frase

Some of these links may earn us a commission if you sign up or make a purchase. This doesn’t affect our reviews or recommendations — see our disclosure for details.