IQ Tester AI is a cognitive assessment agent that delivers a 30-question mixed-stream IQ estimation test across seven cognitive domains and a formal scored report. Unlike generic chatbots that lose domain coverage mid-session, this agent tracks per-domain performance throughout. Built on Jenova's unified platform, it switches models between item writing and report prose across OpenAI, Anthropic, Google, and more. Results are approximate, not a clinically validated IQ test, and are unsuitable for diagnostic, educational, or employment use.
IQ Tester presents one unlabeled question at a time, drawn from Verbal Reasoning, Logical Reasoning, Pattern Recognition, Quantitative Reasoning, Spatial Reasoning, Working Memory, and Abstract Reasoning. Domains are interleaved so the test-taker cannot tell which construct is being measured. Item types include analogies, syllogisms, numerical sequences, mental arithmetic, text-described mental folding, reverse-sequence recall, and invented-rule application — each with a single objectively correct answer.
IQ Tester starts at moderate difficulty and then skews the next item in that domain harder after a correct answer and easier after an incorrect one. Strength on vocabulary depth does not raise cube-counting difficulty, and a miss on a nested conditional does not ease a pattern-breaking sequence. Calibration is never announced; the session converges toward roughly 50% accuracy per domain.
IQ Tester produces a Cognitive Assessment Report only after all 30 answers are scored — never from an abandoned session. The report includes estimated IQ, classification, percentile rank, 95% confidence interval, a dimensional breakdown, raw accuracy and average difficulty reached per domain, a performance narrative, and relative strengths and weaknesses. Tone follows a fixed gradient: flat and clinical below 85, dry through the average range, and more detailed at superior and very superior scores, without consolation or comparison to named individuals.
IQ Tester designs questions using principles drawn from the Wechsler Adult Intelligence Scale, Cattell-Horn-Carroll theory, Raven's Progressive Matrices, and the Stanford-Binet Intelligence Scales. Items target reasoning over knowledge: no advanced physics, niche trivia, or culture-specific idioms. Spatial Reasoning and Working Memory receive fewer items and wider confidence intervals because text chat cannot present figure matrices and cannot prevent scroll-back; those limits are stated in the report's weighting note.
| Traditional Approach | IQ Tester |
|---|---|
| Trivia mixed into supposed IQ quizzes | Reasoning items; no specialized knowledge required |
| Hints, praise, and running score totals | No feedback until all 30 questions |
| Early answers dropped before final scoring | Full-session domain tallies and weighting |
| Partial quizzes still receive a score | No report from abandoned sessions |
Unlimited Memory vs. Chatbot Amnesia
A generic chatbot that drops earlier items cannot keep questions-asked versus target counts (5–6 verbal, 2–3 spatial, 3–4 working memory). It then clusters remaining items by type — revealing the construct — or under-samples a domain and still prints a dimensional score as if the data existed. IQ Tester retains domain allocation, correctness, and per-domain difficulty through all 30 items so the composite score and the domain table describe the same session.
Multi-Model AI Access
Writing a five-premise constraint-satisfaction item or a culturally neutral dual-blank completion rewards a different model profile than writing a Performance Narrative that must stay clinical at a low estimate and become dimensionally specific above 130. Paid tiers can select among OpenAI, Anthropic, Google, xAI, DeepSeek, Qwen, and more, rather than remaining locked to one vendor (ChatGPT is OpenAI-only; Claude is Anthropic-only). Item construction and report prose can therefore use different providers inside one assessment.
Intelligent Tool Integration for End-to-End Workflows
After the 30th answer, the path is scored responses → weighted composite and domain table → in-chat Cognitive Assessment Report → optional PDF or DOCX file, CSV of raw accuracy, a bell-curve image marking estimated IQ on a mean-100 / SD-15 distribution, or Gmail delivery. That chain turns a scored session into an archival document instead of a transcript that vanishes when the chat closes.
Psychometric Item Discipline
Generic assistants pad quizzes with trivia, accept subjective answers, and coach the user. IQ Tester requires one objectively correct response, refuses specialized-knowledge items, scores “I don’t know” as incorrect, and stays silent between questions — including when the test-taker asks how they are doing.
Adults seeking a structured IQ estimate
This agent is for people who want fluid reasoning, verbal deduction, and pattern items rather than general-knowledge quizzes. The session stays unlabeled and uncommented until the report.
People who want a domain-level cognitive profile
The dimensional breakdown separates Verbal Reasoning from Quantitative Reasoning, Working Memory, and Abstract Reasoning so a high composite is not treated as evidence of equal strength in every domain.
Test-takers who want unsparing results
Below-average and low-average estimates are stated as data. The AI does not reframe, console, or debate the numbers if the score is challenged.
Anyone weighing a professional evaluation
Use the output as a private snapshot. School placement, diagnosis, and employment decisions still require a licensed psychologist or psychometrician.
Free For exploring Jenova:
Plus ($20/month) For individuals:
Premium ($50/month) For regular users:
Pro ($100/month) For power users:
Max ($200/month) For professionals:
Ultra ($500/month) For heavy professionals:
Enterprise ($1,000/month) For businesses:
All plans are billed monthly with no daily usage caps — your full quota is available immediately on each billing date.
IQ Tester is an AI administrator of a structured cognitive estimation exercise that presents mixed-domain questions one at a time and then issues a scored Cognitive Assessment Report. It samples seven domains with no mid-test feedback, no item labels, and no running totals. The output is an estimated IQ with percentile rank and domain tables, not a professional psychological evaluation.
IQ Tester retains every item’s domain, difficulty, and correctness so the composite score matches the questions actually delivered; a general chatbot that loses early answers cannot produce a coherent dimensional breakdown. Item writing and report prose can be routed across OpenAI, Anthropic, Google, xAI, DeepSeek, Qwen, and other providers, which matters when a nested-conditional item and a clinical Performance Narrative need different model behavior. After the 30th answer, the same session can export a PDF or DOCX report, a CSV of raw domain accuracy, a marked bell-curve image, or an email via Gmail.
IQ Tester exports the finished Cognitive Assessment Report as PDF or DOCX, writes raw domain counts to CSV or plain text, generates a clinical bell-curve image locating estimated IQ on a mean-100 distribution, and can send those files through Gmail. Files are offered only after the in-chat report exists. An incomplete session produces no document.
No. IQ Tester is a text-based cognitive estimation exercise, not a standardized instrument given by a licensed psychologist. Results are approximate and are not valid for diagnosis, educational placement, or employment decisions. The report repeats this disclaimer in the header and footer.
No. IQ Tester never issues clinical diagnoses, including giftedness, cognitive disability, or learning disorders. A label such as Very Superior describes estimated score range only. Anyone seeking a diagnosis needs a licensed psychologist or psychometrician.
The platform cannot display figure matrices, block-design boards, or Raven-style images, and it cannot measure latency between messages. Spatial items are therefore described in words (folds, painted cubes, compass routes), and speed is not scored. Those constraints widen confidence intervals for Spatial Reasoning and Working Memory and are disclosed in the report’s weighting note.
No. Hints are refused, previous answers are final, and “I don’t know” is scored as incorrect. Off-topic replies are also scored as incorrect. The next unlabeled question follows without commentary.
No assessment is generated from an incomplete session. The agent asks whether to continue or abandon; abandoning produces no partial composite, no percentile, and no domain table. A new session must restart the full 30-item sequence.
No. Correct answers, solution keys, and item-level marks are never disclosed during or after the test. The report shows domain-level accuracy and average difficulty reached, not which specific analogy or syllogism was missed.
Yes. A retake uses a different item set and resets scoring state, though familiarity with formats may reduce comparability. Prior results remain visible in the chat history above the new assessment.
Yes. IQ Tester runs with full feature parity on web, iOS, and Android, including the 30-question session and report export. The iOS app is at https://apps.apple.com/app/jenova/id6747692301 and the Android app is at https://play.google.com/store/apps/details?id=ai.jenova.app.prod.
No. Conversations and scores are not used to train public models, are encrypted in transit and at rest, and are not sold or shared with advertisers. Treat the report as personal estimation data, not a clinical record.
IQ Tester is part of Jenova, the most powerful AI agent platform in the world. Jenova provides unified access to leading AI models from OpenAI, Anthropic, Google, xAI, DeepSeek, Qwen, and more — alongside intelligent tool integrations for end-to-end task completion. The platform features persistent cross-session memory, automated workflow orchestration, and full parity across web, iOS, and Android. Developed by Azeroth Inc., a New York-based technology company, Jenova serves users in over 70 countries.
For developers, the Jenova Agent API provides a fully-managed agent backend with built-in orchestration, multi-model routing, memory, storage, and MCP tool support — replacing the need to stitch together raw LLM APIs, vector databases, and custom tool servers.
Contact: [email protected]