Top 5 Best Online IQ Tests
A Comprehensive Comparison

0 The short answer

Reviewed September 25, 2026 by Structural. The best online IQ test depends on the result you need. Compare ACIS and RIOT when you want a paid assessment with published technical material; examine CORE when a free, broad battery is your priority; consider AGCT for a shorter, historically derived format. CAIT remains here because readers encounter it, but its publisher now directs new users to CORE.

A useful choice starts with purpose, eligibility, evidence and practical requirements. No overall point total can make a free exploratory battery, a paid report and an assessment accepted by a particular institution interchangeable. This page therefore shows the basis and limits of each recommendation instead of claiming that one brand wins for everyone.

Jump to the five-option overview, the review method, the other formats or the frequently asked questions. Product-specific sources sit beside the claims they support.

1 How was this comparison checked?

We reviewed public primary sources on September 25, 2026. Those sources include the ACIS technical manual, RIOT's assessment catalogue and manual overview, and CognitiveMetrics' product and technical pages. A source can establish what a publisher currently discloses. Reading it does not independently reproduce the publisher's analyses or prove that every user receives the advertised experience.

The comparison asks the same practical questions of each product: What tasks contribute to the result? Who is the reference population? Which version does the evidence describe? What does the report include? What must the reader pay or complete? Which security and retest procedures are documented? Is the score suitable for the reader's intended use? An unanswered question remains unknown; it does not become a zero.

We do not use the previous numerical quality totals. Assigning equally weighted points to security, precision, clinical relevance and interface features implied a common measurement scale that this review cannot substantiate. A narrower test can be a sensible choice for a narrower question, and an openly stated limitation can be more useful than a confident marketing promise.

We did not purchase competing products, inspect their protected item banks, audit their scoring code or compare the same participants across batteries. Prices mentioned here are public page observations, not completed checkout quotes. Source dates and our review date are separate: an older technical report can remain the publisher's current document, while a newly redesigned product page may provide little technical evidence. Those distinctions matter more than a fresh-looking badge.

2 What do the five options offer?

The following overview is a starting point for reading, not a league table. Each product's section explains its evidence and the decision it might serve. Paid access, free access and a legacy status describe availability; none is a scientific quality grade. Follow the linked primary material before relying on a feature that is essential to your decision.

These are five options selected for a focused comparison, not an exhaustive census of every test online. Their inclusion does not establish institutional acceptance or equal depth of evidence. In particular, historical use of an underlying test and present-day validation of a web implementation are different questions. A brand's largest sample headline may also refer to all visitors, partial completions or a different test version.

For the reader choosing today, compare the exact configuration available to you. A short form and a full battery under the same brand can produce different information. The general comprehensive IQ test guide explains coverage; the nonverbal IQ test guide addresses reduced verbal demand. Neither selection should be made by brand position in a list.

3 When might ACIS fit your goal?

Documented: The ACIS technical manual describes a Full Scale assessment with 20 subtests across six primary domains, along with composite and subtest reporting. It presents methods for norming, reliability, uncertainty and interpretation. These are ACIS's own technical disclosures. Their public availability helps a reader inspect the rationale, but is not an independent endorsement of the product.

Current limitation: The manual's revision notice keeps version 1.4 as the reference while updates are prepared. It records the retirement of Spatial Navigation on September 7, 2026 and the introduction of Layer Rotation; revised norms and dependent tables are pending the next edition. Historical values for the retired task should not be read as validation of its replacement. This comparison does not fill that gap with estimated figures.

Editorial fit: ACIS is worth examining when you want to look beyond a single total and understand how a completed battery is organized. The relevant purchasing question is whether the selected assessment and its report answer your purpose. A reader who wants broad coverage should inspect the included domains; a reader interested in one ability should avoid assuming that more tasks automatically make a better purchase.

Read a sample report before testing so that the score hierarchy and interpretation language are familiar. The guide to detailed IQ results provides questions for checking whether a report explains its scores. A report can contain many pages without answering a reader's central question, so evaluate clarity as well as the number of outputs.

Not independently verified here: superiority over the other four tests, clinical equivalence, institutional acceptance, and the comparative effectiveness of security controls. ACIS has the same burden of evidence as the alternatives. Publication by ACIS creates a commercial interest that a reader should keep in view even when an individual statement can be traced to the manual. The related reliability and validity explanation distinguishes the questions those terms address.

Reading ACIS during a documented revision

The forms section and interpretation rules provide a more useful starting point than the largest number on a promotional page. Start with the form you plan to complete, identify which outputs it supports, and then read the technical discussion for those outputs. If your main interest is a particular domain, check its component tasks rather than treating the Full Scale label as a promise about every possible interpretation.

The active revision notice makes that exercise especially relevant for visual spatial interpretation. A prospective reader should distinguish three things: the task currently delivered, the evidence retained for a previous task, and the evidence promised for the next edition. Those are different states of documentation. Leaving them separate gives the reader an honest account of what is available today. It would be inappropriate to copy an old reliability value into a claim about the replacement task because the surrounding assessment still has the same name.

For comparison with another product, first match the kind of question being asked. A total-score question, a broad-domain question and a detailed subtest question need different support. ACIS's public documentation makes these questions inspectable, while the pending revision remains a reason to qualify interpretations that depend on the affected material.

4 What is documented about RIOT?

Observed: RIOT's assessment catalogue lists a Full configuration with 15 subtests and a Basic configuration with five. At review, it displayed promotional prices of $50 for Full and $25 for Basic. These are page observations in the displayed currency, not a guarantee of the final price, promotion duration or regional checkout terms.

Publisher-documented: The RIOT manual overview identifies version 1.0, publication in 2025 and an adult age range starting at 18. It describes technical coverage and offers a preview and a route to request the full manual. We reviewed that public overview, not the complete restricted document. A linked report example was available, but its interactive contents were not fully inspectable in this review.

Editorial fit: RIOT merits a closer look when a paid package with a defined selection of tasks fits your budget and intended time commitment. The meaningful comparison is between the specific package you would buy and the alternatives you would actually complete. Do not compare a competitor's brief form with a full battery and attribute the difference to brand quality.

Before purchasing, check whether the report detail you want is included in the selected package. If subtest interpretation matters, ask what remains visible after completion and what explanations accompany it. If documentation depth matters, request the technical material needed to evaluate your specific use. A request-access process is neither proof of missing evidence nor a substitute for reading the evidence.

Not independently verified here: the complete manual's analyses, actual completion times across users, checkout experience, and direct score agreement with other products. RIOT's assessment page and manual overview provide different time estimates for full testing, so we do not award a speed advantage from either figure. Allow for instructions and breaks, and verify the estimate for the configuration you select. A convenient format matters only if the result still addresses the question that brought you to the test.

Choosing a RIOT package by its actual output

The Basic product page names Vocabulary, Matrix Reasoning, SToVeS, Visual Reversal and Symbol Search. That is more informative than simply calling the package quick. A prospective buyer can compare those tasks with the abilities they wanted to investigate and decide whether the omissions matter. The Full product page advertises global, index and subtest outputs, so report depth is a concrete difference to investigate between the packages.

The Full page also distinguishes its stated norm population from comparisons with other RIOT users in the report. Those are potentially different reference groups. As a reader, ask which comparison a particular graph is displaying: standing in the norm group or standing among people who chose to use the website. The second group should not silently be treated as the first. This is an interpretation question, not an allegation that either display is incorrectly calculated.

RIOT's published material therefore offers specific questions to resolve before purchase. Which output is included in your package? Which population does it use? Which technical table supports that score? Answering these is more useful than awarding a point for the mere existence of a manual, and it gives RIOT credit for the concrete information it does disclose.

5 What makes CORE worth considering?

Observed: CognitiveMetrics presents CORE as a free, community-developed battery with broad CHC coverage. The overview describes abbreviated, optimized and full testing paths, with different time commitments. It also discusses language suitability. Readers should choose a path intentionally rather than assuming that every result under the CORE name reflects the same completed tasks.

Publisher-documented: CORE's preliminary technical report, version 0.2 dated December 21, 2025, includes analyses and limitations. It describes a self-selected sample interested in cognitive testing and procedures for excluding invalid attempts. That is concrete documentation to inspect. It does not justify treating the sample as a representative general population or claiming that its security controls have independently demonstrated superiority.

The separate test structure page explains task design and differences from tests that inspired it. For example, it documents different processing-speed options for touchscreen and computer users. This helps identify a practical requirement before starting. The publisher's rationale for an adaptation should still be distinguished from an independent study showing equivalence between administrations.

Editorial fit: CORE is a relevant option when cost is a binding constraint and you are willing to read technical qualifications. Free access does not mean an absence of science, just as payment does not establish validity. Its public reporting makes it possible to ask specific questions about interpretation instead of judging the product from its price category.

Not independently verified here: reproduction of its analyses, how well its sample represents a particular reader, institutional acceptance, and security effectiveness in use. For an exploratory self-assessment, you may accept limits that would be unacceptable for a consequential decision. State that purpose before interpreting the result. Our Cognitive Metrics review offers additional navigation, but the current publisher pages linked here are the source for the product facts in this section. Documentation should be read at the level of the score and testing path you intend to use.

CORE's flexibility creates a configuration decision

The CORE overview specifies Analogies, Figure Weights and Digit Span for the abbreviated path. It describes an optimized path below two hours and a full path of approximately three hours, with the latter possible across sessions. These are publisher estimates. The practical distinction is that choosing selected tasks and completing the full battery are different commitments with different output scope.

The structure documentation also explains an auditory Digit Span format and a computer-based Character Pairing option alongside touchscreen Symbol Search. A reader planning to use a laptop should inspect the intended computer path before starting a speed task. A reader with difficulty hearing audio should resolve the administration requirements rather than substituting a personally convenient presentation and assuming the score remains comparable.

These details make CORE more than a generic free alternative, but they also make preparation important. Write down which configuration you chose and use the interpretation attached to that configuration. Flexibility is valuable when it helps a reader select a suitable assessment; it can confuse comparisons if two people both say they took CORE while completing substantially different sets of tasks.

Its native-English guidance is another selection constraint. A broad menu does not automatically solve the language requirements of every included score. Consider the publisher's guidance for the score you intend to interpret, and do not treat a label about reduced cultural demand as proof that all population differences have been removed.

6 What should you check before taking AGCT?

Observed: The current CognitiveMetrics AGCT page offers forms 1a and 1c, describes verbal, quantitative and spatial content, and advertises a 40-minute test. It states that a $10 payment after testing unlocks the full score profile. That announced fee belongs in a buying comparison; it should not be hidden behind a broad statement that all community tests are free.

Evidence boundary: The publisher relates the test to the historical Army General Classification Test. Evidence concerning an original instrument can inform a review, but cannot by itself establish that every present-day browser adaptation has equivalent administration, scoring or norms. The relevant questions are which form is delivered now, which conversion is applied, and what evidence connects the current implementation to its intended interpretation.

Editorial fit: A reader looking for a more compact task set may prefer a clearly bounded sitting over a longer battery. That preference is legitimate. It also changes the information available afterward: select the format because its scope suits your question, rather than assuming that a concise result reveals every cognitive domain. Ask which components contribute to the total and which do not.

Read the result-access statement before committing the testing time. Separate the ability to begin the test from the ability to view the output you want. If you are comparing products on cost, compare the required payment for the desired result, not simply whether the landing page offers a start button. The free versus validated IQ tests guide explains why access model and evidence quality must be considered separately.

Not independently verified here: checkout delivery, the adequacy of current norms for every population, and comparative precision against broader batteries. We do not turn the historical background into a modern endorsement, nor penalize the product merely because its design is older. A narrower, well-understood purpose can make a test useful; the unsupported step would be extending that result to decisions the current evidence has not addressed.

Form selection is part of the AGCT decision

The current AGCT description recommends form 1c for a first attempt and explains that its historical development addressed limitations of the earlier form. That is the publisher's recommendation, not a new finding from this review. It provides a practical reason to pay attention to the form identifier instead of following whichever start link appears first in an older recommendation.

If you already have an AGCT result, record its form before comparing it with a new one. A difference between results might reflect more than a change in your performance, so first establish whether the tasks, conversion and conditions were the same. For a new attempt, ask whether the desired profile will answer your original question after the announced fee. Choosing a compact assessment knowingly is different from expecting a complete domain profile because a historical test name sounds comprehensive.

7 Why is CAIT still listed if CORE replaced it?

Observed: The CAIT page now says that CORE has replaced the test. Its validity tab is marked as being in revision. The page still provides information about CAIT's subtests and score calculation and presents it as an informative resource, not a substitute for professional assessment. Those current notices matter more to a new reader than an old ranking position.

Editorial fit: CAIT remains relevant when someone is trying to understand an earlier CAIT result or a recommendation written before CORE became the publisher's preferred option. A new user should first compare the current successor with other current products. This is a maintenance and purpose judgment, not a claim that all previous CAIT results are worthless or that every newer test is automatically better.

A replacement notice should trigger several questions. Does an older score refer to the same tasks and norms still described today? Has the interpretation guidance changed? Is a link in an older review pointing to the intended version? Keep the result date and test name together when discussing a past administration. Treat a renamed or redesigned product as a version question until the publisher explains what changed.

Similarity to a familiar clinical task does not make a community battery an official version of that instrument. The same caution applies to every test on this page. Readers who need a named professional assessment should investigate that exact assessment and the requirements of the organization that will use it. The professional versus online assessment guide explains the practical distinction without assuming that delivery through a browser settles quality.

Not independently verified here: current comparative accuracy, the status of every historical norm table, and continuity of all older reports. We do not manufacture a low quality score from those unknowns. The appropriate action is to identify the version and use the publisher's current guidance. A legacy product can still be informative to past users while being a less straightforward starting point for someone selecting a first assessment today.

Using a legacy calculator carefully

The CAIT page retains a calculator with fields for component scores. A calculator is a tool for combining inputs under its specified rules; its existence does not independently establish the quality of the inputs or make a different test's scores interchangeable. If you are reconstructing an older report, identify what each field expects before entering a number copied from elsewhere.

This is where a legacy result can require a different decision from a new purchase. A past user may need to preserve the original score, understand the old index labels and document what was actually completed. A new user needs to choose a currently supported starting point. Sending both readers to the same recommendation without explaining that distinction would be unhelpful. Our inclusion of CAIT serves the first reader as well as explaining the publisher's current direction to the second.

8 How do other online formats change the comparison?

Other products can serve different purposes without fitting neatly into this five-option overview. The useful distinction is the task, output and intended use, rather than whether a site looks polished or describes itself as scientific. A reasoning exercise, an adaptive estimate and a broad battery should be compared on the question each is designed to answer.

For example, BRGHT's current product page describes an adaptive format and reports numerical, spatial and logical reasoning results. It also permits repeated testing. These are publisher statements about its format, not evidence from our own administration. We have not verified its complete technical model or measured agreement with the five products above. The BRGHT review gives readers a separate place to examine that product rather than attaching an invented point total here.

The Mensa Norway online test is another familiar destination. This review does not establish that completing that page qualifies a person for membership or meets any organization's admission requirements. A recognizable name should prompt verification of the exact service being offered, especially when an organization operates both public exercises and supervised assessments.

Readers investigating commercial result pages can use the MyIQ review as a separate starting point. We are not assigning MyIQ a comparative score here because this update did not complete a current review of its purchase flow and technical evidence. That missing verification is a limit of this article, not proof of the product's quality in either direction.

For any alternative, write down what you need to learn before evaluating it. If you want to explore one type of puzzle, broad profile reporting may add little to that goal. If you want to understand several domains, one task format may leave important questions unanswered. If you need an externally accepted result, acceptance requirements take priority over convenience. These different aims explain why a single best-to-worst list can mislead even when every product description is accurate.

Do not treat a longer list of alternatives as better advice by itself. A useful shortlist should reduce uncertainty about the next step. If two products meet your requirements, compare the particular unresolved issue, such as report access or language suitability, instead of collecting more brand names.

9 Which option fits the decision you are making?

Choose the purpose before choosing the product. A personal exploration, a research administration and an externally required assessment impose different requirements. Start by writing one sentence describing the decision the result will inform. That sentence makes it easier to reject an attractive feature that does not actually help, and to identify a missing feature that matters.

For personal exploration, consider whether you want a broad profile or a more focused estimate. Inspect the exact score outputs and the explanation offered for them. Someone curious about relative strengths may want several domains; someone with limited available time may prefer a narrower task set and accept its narrower scope. Neither choice needs to be defended as universally superior. Completion under appropriate conditions is more useful than buying a larger package you cannot reasonably finish.

When another organization will receive the result, ask that organization for its accepted assessments, administration requirements and reporting rules before paying. This article has not established that any of the five products meets a particular employer's, university's or membership body's requirements. A screenshot with a high number does not answer those procedural questions. Keep the intended recipient's written requirements with your selection notes.

For research or organized testing, administration is part of the purchase. You may need participant management, consent handling, exports, support and a documented scope of permitted use. The cognitive assessment platforms guide and platform cost guide address that separate decision. The existing ACIS Professional information is relevant to that use case; it should not be confused with an individual consumer report.

Set a small number of non-negotiable requirements, such as the appropriate age range, an accessible testing setup and a report you can understand. Then list preferences, such as visual presentation or optional extra domains. Meeting a preference cannot compensate for failing a requirement. This makes comparison manageable without pretending that every feature belongs on one numerical scale.

If no shortlisted option meets a requirement, treat that as an unresolved selection problem. Request the missing information or choose a different route. Lowering the requirement silently because a test is convenient can leave you with a completed assessment that is unusable for the decision you originally wanted to make.

Two examples of a defensible shortlist

Consider a reader who wants a broad personal exploration and has no budget for a paid result. Within this article, CORE is an obvious option to investigate first because its public access model meets that constraint. The next checks are its language guidance, selected testing path, required equipment and technical limitations. Paying for a different brand would not be a necessary first step merely because it appears earlier on the page.

Now consider a reader willing to pay who wants a report they can examine before choosing. ACIS and RIOT both offer report-related material, but this review could inspect different amounts of that material. The reader should open the actual example available to them and check the outputs of the package they would select. If a crucial page does not load, that is an unresolved information need, not a reason to assume the report is either excellent or absent.

Neither example produces a universal winner. Each starts from a real constraint and narrows the next verification. A different reader whose institution requires a named assessment may reject the whole shortlist immediately, even if every product here is convenient and affordable.

10 How should you compare price, time and completion?

Compare the cost of the result you intend to use. A free start, a free total, a paid profile and an included report are different offers. This article records selected public price statements alongside their sources, but does not guarantee a final checkout amount. Confirm the current currency, required payment, included output and access conditions on the product's own page before committing.

Separate testing time from total time. Instructions, equipment checks, breaks and reading the report all require attention, even when a publisher's headline refers only to answering questions. A short form can be useful when time is limited, but its score should be interpreted according to that form's evidence. Do not borrow the evidence or breadth of a longer version simply because both share a brand.

Think about the session you can actually arrange. Do you have the required input device? Can you hear any audio clearly? Are the instructions understandable in your language? Can you avoid interruptions for the required sitting? The appropriate answers depend on the test's published administration rules. Confirm whether breaks are permitted, when they are permitted and whether leaving a session affects the result.

Before you pay, identify what happens if the session is interrupted. Check the provider's stated policy for resuming, retesting and accessing support. This review has not tested every recovery path, so it makes no cross-product claim about which handles interruptions best. A policy visible on a page and a successfully reproduced recovery are different forms of evidence. If uninterrupted completion is difficult for you, resolve that question before starting.

When comparing value, ask which part of a larger package you would use. More outputs can help when they answer a real question, but they can also add reading and decision time. Conversely, a lower initial fee may not be economical if the information you need requires an additional purchase. Record the offer you chose and the date, especially if a sale influenced the decision.

A practical comparison note needs only a few fields: selected form, required conditions, expected commitment, included result, current payment and unresolved question. If one of those fields is unknown, keep it visibly unknown. This small step prevents a polished feature list from becoming an assumption about what you will receive after completion.

11 What evidence makes a score interpretable?

Begin with the meaning of the score rather than the number of claims on the sales page. A technical document should let you identify the construct, the scoring method, the reference group and the scope of the proposed interpretation. If the same product offers several forms, locate the evidence for the form and score you will receive. Evidence for a total does not automatically explain every subtest.

Norms provide a comparison with a reference population. Ask who entered that population, when data were collected, how participants were selected and which restrictions apply. The guide to IQ score norming explains those questions in more detail. A large sample label alone does not tell you whether the sample represents your age, language background or testing conditions.

Reliability concerns consistency under specified conditions; validity concerns the support for an interpretation and use. Neither is a universal quality certificate attached to a brand. Read the method behind any coefficient and check which score it describes. When two publishers use different samples, models or score definitions, placing their headline coefficients next to each other can create an unfair appearance of precision.

Uncertainty should also be visible in interpretation. A confidence interval is meaningful only with its stated method and assumptions, and a percentile is a position within a reference distribution. Neither turns a score into a complete description of a person. If a report presents a total and domain differences, ask what guidance supports discussing those differences. Do not infer a diagnosis or a life outcome from a pattern simply because the chart is detailed.

Our online IQ test accuracy guide separates questions about the web format from questions about measurement evidence. Browser delivery alone does not settle accuracy. Equally, familiarity with a task name is not enough to establish that a new implementation has the same properties as an established instrument.

For a careful reading, trace one important claim all the way to its supporting passage. Confirm the version, sample, method and limitation before moving to the next claim. If the source does not answer the question, the responsible conclusion is that this review cannot establish it. That is more informative than filling a comparison cell with an apparently exact rating derived from intuition.

12 What can this review say about security and retakes?

It can identify documented procedures and clearly state the limits of verification. It cannot assign a trustworthy comparative security score without a common audit protocol. A website may describe attempt screening, retake restrictions or protected delivery. Those descriptions are relevant, but they do not demonstrate how effectively each system detects outside help or prevents prior item exposure.

CORE's technical report describes excluding attempts affected by outside assistance or insufficient engagement. The ACIS manual's security section describes integrity and retest procedures. These are publisher-documented controls. This update does not reproduce their detection methods, establish false-positive rates or show that one product is more resistant to misuse than another.

For your own result, follow the provider's instructions on permitted materials and assistance. Do not assume that a calculator, notes, another person's help or a separate device is allowed simply because the browser does not stop you from using it. The point of a personal assessment is to obtain an interpretable account of your performance under the intended conditions, not to discover a way to increase the displayed number.

Retesting raises a separate interpretation question. Keep track of prior exposure, the interval between attempts and whether the task or form changed. A policy allowing another attempt is a product feature; it does not by itself explain how comparable the later score is to a first attempt. Read the provider's retest guidance before deciding which result to retain or share.

Security can also affect the reader's experience. A restriction may support standardization while making testing less convenient. Evaluate whether you can meet the conditions and what help is available if a legitimate interruption occurs. Do not infer a defect from an access restriction without understanding its purpose, or infer proven protection from a visible warning banner.

If you are selecting a test for a group, ask for documentation appropriate to that use. Clarify identity requirements, supervision, handling of irregular sessions and score review procedures. This article does not certify any product for consequential selection. A sensible evaluation separates advertised controls, documented methods, observed operation and independently measured effectiveness, instead of collapsing them into one security star rating.

13 Where can you check the evidence and decide your next step?

Use the primary links in each product section to verify the claims that matter to you. They support the descriptions on this page as reviewed on September 25, 2026. A linked source is not an endorsement of every statement on the publisher's website, and a later revision may change a feature or limitation. Keep the exact form and source date together when comparing notes.

For background, the following retained references help readers investigate testing, score reporting and interpretation. They are not studies comparing the five products in this article. Their presence must not be read as evidence that a professional publisher, independent review organization or research author has endorsed ACIS or any other listed product.

• American Psychological Association: understanding psychological testing and assessment.
• Buros Center for Testing, a starting point for investigating independent test reviews.
• Pearson's WAIS-5 product information and Riverside's Stanford-Binet product information, for readers investigating those specific professional instruments.
• Pearson's WAIS-IV sample score report and clinical assessment primer, as background reading about reporting.
• Voncken and colleagues on confidence intervals for normed scores, Crawford and colleagues on percentile norms, and Trahan and colleagues' Flynn effect meta-analysis, retained as general research references rather than product validation.

If ACIS fits your requirements, inspect the selected assessment and report before committing. If another product fits better, use its primary documentation to resolve the remaining question. The useful outcome of this guide is a choice you can explain in terms of purpose and evidence, with any remaining uncertainty still visible.

The general evaluation framework is the 2014 Standards for Educational and Psychological Testing, jointly published by AERA, APA and NCME, with an additional reference to APA's testing standards information. These standards are a framework for examining evidence and use, not a certification granted to the products in this comparison.

14 Frequently asked questions