[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"$f1xa9n3du7od3v":3,"$f27qo82yht9ynw":95},{"success":4,"data":5},true,{"id":6,"slug":7,"coverImage":8,"author":9,"viewsCount":10,"createdAt":11,"title":12,"description":13,"lang":14,"contentHtml":15,"readingTime":16,"toc":17},"u65iu8mijuzx0t4g1981phue","iq-deepseek-v4","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1677372139014-394e5b51f79b?w=1280","Dr. Daria Dudnik",0,"2026-08-01T09:20:10.570Z","What is the IQ of DeepSeek V4?","DeepSeek V4 Pro scores 111 on the Mensa Norway IQ test and 44.3 on the Intelligence Index. We break down every benchmark and explain why it's the best budget AI model.","en","\u003Ch2 id=\"deepseek-v4-pro-china-s-open-source-champion\">DeepSeek V4 Pro: China's open-source champion\u003C\u002Fh2>\n\u003Cp>DeepSeek V4 Pro is the flagship model from DeepSeek, the Chinese AI lab that shocked the world with its combination of low cost and high performance. Released in early 2026, DeepSeek V4 Pro scores \u003Cstrong>111\u003C\u002Fstrong> on the Mensa Norway IQ test and \u003Cstrong>44.3\u003C\u002Fstrong> on the Artificial Analysis Intelligence Index v4.1 — placing it in the top 10 globally, but well behind the frontier models (Claude at 55.7, GPT-5.5 at 54.8).\u003C\u002Fp>\n\u003Cp>What makes DeepSeek V4 Pro remarkable is not its raw intelligence — it's its \u003Cstrong>price\u003C\u002Fstrong>. At 0.44\u002F0.87 per 1M tokens, it costs roughly \u003Cstrong>10x less\u003C\u002Fstrong> than Claude (5\u002F25) and GPT-5.5 (5\u002F30), and \u003Cstrong>5x less\u003C\u002Fstrong> than Gemini (2\u002F12). For the price of a single query to Claude, you could run ten queries to DeepSeek V4 Pro. And yet, it scores 111 on an IQ test — higher than approximately 25% of humans.\u003C\u002Fp>\n\u003Cp>DeepSeek V4 Pro is the \u003Cstrong>people's AI\u003C\u002Fstrong>: a model that doesn't compete at the frontier of intelligence, but brings capable AI to everyone at a price that's practically free. For students, hobbyists, small businesses, and developers in cost-sensitive markets, DeepSeek V4 Pro is often the only viable option.\u003C\u002Fp>\n\u003Cp>\u003Cstrong>Key scores:\u003C\u002Fstrong>\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>Artificial Analysis Intelligence Index\u003C\u002Fstrong>: 44.3 (top 10 globally)\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Mensa Norway IQ (TrackingAI)\u003C\u002Fstrong>: 111 (above average human)\u003C\u002Fli>\n\u003Cli>\u003Cstrong>AI IQ composite (VexoWire)\u003C\u002Fstrong>: ~111 estimated\u003C\u002Fli>\n\u003Cli>\u003Cstrong>GDPval-AA v2\u003C\u002Fstrong>: top 10\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Humanity's Last Exam\u003C\u002Fstrong>: top 10\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Hallucination rate\u003C\u002Fstrong>: moderate\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Cost\u003C\u002Fstrong>: 0.44\u002F0.87 per 1M tokens (lowest among all major models)\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Open-source\u003C\u002Fstrong>: weights available for local deployment\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Chr>\n\u003Ch2 id=\"1-the-budget-revolution\">1. The budget revolution\u003C\u002Fh2>\n\u003Cp>DeepSeek V4 Pro represents a fundamental shift in the AI landscape: the democratization of intelligence. While American labs (OpenAI, Anthropic, Google, xAI) compete at the frontier — pushing IQ scores from 130 to 140 to 145 — DeepSeek has focused on a different goal: making capable AI affordable to everyone.\u003C\u002Fp>\n\u003Cp>At 0.44 per 1M input tokens and0.87 per 1M output tokens, DeepSeek V4 Pro is staggeringly cheap. To put this in perspective: processing a typical book-length document (about 100,000 tokens) costs approximately 0.04 with input and0.09 with output. The same task would cost 0.50-1.00 with Gemini, 1.00-3.00 with Claude or GPT-5.5.\u003C\u002Fp>\n\u003Cp>This isn't just a discount — it's a different category of product. DeepSeek V4 Pro makes AI accessible to:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>Students\u003C\u002Fstrong> who can't afford $20\u002Fmonth subscriptions\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Small businesses\u003C\u002Fstrong> in developing markets\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Hobbyists and tinkerers\u003C\u002Fstrong> running experiments\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Developers\u003C\u002Fstrong> building high-volume applications\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Researchers\u003C\u002Fstrong> processing massive datasets on limited budgets\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>And the remarkable thing is: DeepSeek V4 Pro is genuinely capable. It scores 111 on the Mensa Norway IQ test — above the human average of 100, and higher than about 25% of people. It's not a genius, but it's smart enough for most everyday tasks: writing, coding, analysis, translation, summarization.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2 id=\"2-benchmark-breakdown\">2. Benchmark breakdown\u003C\u002Fh2>\n\u003Ch3 id=\"mensa-norway-iq-test-111-above-average\">Mensa Norway IQ test: 111 (above average)\u003C\u002Fh3>\n\u003Cp>The Mensa Norway IQ test consists of 36 visual pattern recognition questions. DeepSeek V4 Pro scores \u003Cstrong>111\u003C\u002Fstrong> — above the human average of 100, placing it in the "above average" range. This is higher than approximately 25% of humans, but well below the frontier models (Claude ~130, Gemini 141, GPT-5.5 ~145, Grok-4.20 145).\u003C\u002Fp>\n\u003Cp>An IQ of 111 means DeepSeek V4 Pro can recognize visual patterns and reason abstractly at a level comparable to a bright college student. It won't solve the hardest puzzles, but it handles most everyday reasoning tasks competently.\u003C\u002Fp>\n\u003Cp>The gap between DeepSeek V4 Pro (111) and the frontier models (130-145) is significant — about 20-35 IQ points, or roughly 1-2 standard deviations. This means the frontier models are substantially smarter on raw cognitive tests. But for tasks that don't require genius-level reasoning — which is most tasks — DeepSeek V4 Pro is more than adequate.\u003C\u002Fp>\n\u003Ch3 id=\"artificial-analysis-intelligence-index-v4-1-44-3-top-10\">Artificial Analysis Intelligence Index v4.1: 44.3 (top 10)\u003C\u002Fh3>\n\u003Cp>On the Intelligence Index, DeepSeek V4 Pro scores \u003Cstrong>44.3\u003C\u002Fstrong> — placing it in the top 10 globally, but about 10-11 points behind the frontier leaders (Claude 55.7, GPT-5.5 54.8). This gap is significant but expected given the price difference.\u003C\u002Fp>\n\u003Cp>Breaking down the Index components:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>GDPval-AA v2\u003C\u002Fstrong>: DeepSeek scores reasonably but trails Claude and GPT-5.5 on complex multi-step tasks\u003C\u002Fli>\n\u003Cli>\u003Cstrong>AA-Omniscience\u003C\u002Fstrong>: DeepSeek has a moderate hallucination rate — not as low as Claude, but acceptable for most uses\u003C\u002Fli>\n\u003Cli>\u003Cstrong>GPQA\u003C\u002Fstrong>: DeepSeek performs adequately on graduate-level science questions\u003C\u002Fli>\n\u003Cli>\u003Cstrong>HLE\u003C\u002Fstrong>: DeepSeek ranks in the top 10 on Humanity's Last Exam\u003C\u002Fli>\n\u003Cli>\u003Cstrong>ARC-AGI\u003C\u002Fstrong>: DeepSeek struggles with abstract reasoning compared to frontier models\u003C\u002Fli>\n\u003Cli>\u003Cstrong>LMSYS Arena\u003C\u002Fstrong>: DeepSeek gets decent human preference scores, particularly for coding tasks\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Ch3 id=\"ai-iq-composite-vexowire-111\">AI IQ composite (VexoWire): ~111\u003C\u002Fh3>\n\u003Cp>The VexoWire composite IQ blends multiple tests (Mensa Norway, Mensa Denmark, Raven's, WAIS-IV). DeepSeek V4 Pro's estimated composite IQ is \u003Cstrong>~111\u003C\u002Fstrong> — consistent with its Mensa Norway score. This places it firmly in the "above average" range, comparable to a bright human but well below genius level.\u003C\u002Fp>\n\u003Ch3 id=\"gdpval-aa-v2-top-10\">GDPval-AA v2: top 10\u003C\u002Fh3>\n\u003Cp>On the agentic benchmark, DeepSeek V4 Pro ranks in the top 10. It's significantly less capable than Claude (1,890 Elo) and GPT-5.5 (1,850 Elo) on complex multi-step tasks, but can handle moderate agentic workloads. For simple automation, basic coding tasks, and straightforward multi-step instructions, DeepSeek V4 Pro is competent.\u003C\u002Fp>\n\u003Ch3 id=\"humanity-s-last-exam-hle-top-10\">Humanity's Last Exam (HLE): top 10\u003C\u002Fh3>\n\u003Cp>On the hardest academic questions, DeepSeek V4 Pro ranks in the top 10 among AI models. This reflects DeepSeek's strong training on academic and scientific data — the model has broad knowledge across disciplines, even if it can't match the depth of reasoning of frontier models.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2 id=\"3-what-does-an-iq-of-111-mean\">3. What does an IQ of 111 mean?\u003C\u002Fh2>\n\u003Cp>To put DeepSeek V4 Pro's IQ of 111 in context:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>85-115 (68% of people)\u003C\u002Fstrong>: Average intelligence\u003C\u002Fli>\n\u003Cli>\u003Cstrong>115-130 (14%)\u003C\u002Fstrong>: Above average to high average\u003C\u002Fli>\n\u003Cli>\u003Cstrong>130-145 (2%)\u003C\u002Fstrong>: Gifted — qualifies for Mensa (top 2%)\u003C\u002Fli>\n\u003Cli>\u003Cstrong>145-160 (0.1%)\u003C\u002Fstrong>: Highly gifted \u002F genius\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>DeepSeek V4 Pro, at 111, sits in the \u003Cstrong>upper end of the average range\u003C\u002Fstrong> — smarter than about 25% of humans, but not quite reaching the "above average" threshold of 115. If it were a person, it would be comparable to a bright college student or a capable professional — not a genius, but certainly competent.\u003C\u002Fp>\n\u003Cp>This is notably lower than the frontier models:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>Claude Opus 4.8: ~130 (gifted)\u003C\u002Fli>\n\u003Cli>Gemini 3.1 Pro: 141 (highly gifted)\u003C\u002Fli>\n\u003Cli>GPT-5.5: ~145 (genius)\u003C\u002Fli>\n\u003Cli>Grok-4.20: 145 (genius)\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>The gap is significant — about 20-35 IQ points. But here's the key question: does the gap matter for your use case? For most everyday tasks — writing emails, summarizing documents, basic coding, translation — an IQ of 111 is more than sufficient. The frontier models' advantage shows up on the hardest problems: complex mathematical proofs, intricate multi-step reasoning, cutting-edge research questions.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2 id=\"4-where-deepseek-v4-pro-excels\">4. Where DeepSeek V4 Pro excels\u003C\u002Fh2>\n\u003Ch3 id=\"cost-effectiveness\">Cost-effectiveness\u003C\u002Fh3>\n\u003Cp>This is DeepSeek V4 Pro's defining feature. At 0.44\u002F0.87 per 1M tokens, it's by far the cheapest major AI model. For high-volume applications — chatbots serving thousands of users, bulk document processing, large-scale data analysis — the cost savings are enormous. You could run 10-30 DeepSeek queries for the price of a single Claude or GPT-5.5 query.\u003C\u002Fp>\n\u003Ch3 id=\"coding-and-technical-tasks\">Coding and technical tasks\u003C\u002Fh3>\n\u003Cp>DeepSeek V4 Pro is particularly strong at coding tasks. DeepSeek has invested heavily in code training data, and the model performs well on benchmarks like HumanEval and MBPP. For developers who need a coding assistant but can't afford frontier prices, DeepSeek V4 Pro is an excellent choice.\u003C\u002Fp>\n\u003Ch3 id=\"open-source-availability\">Open-source availability\u003C\u002Fh3>\n\u003Cp>DeepSeek V4 Pro's weights are available for local deployment. This means organizations can run the model on their own hardware, avoiding API costs entirely and keeping data private. For security-conscious organizations, researchers in restricted environments, and developers who need full control, this is a major advantage.\u003C\u002Fp>\n\u003Ch3 id=\"multilingual-support\">Multilingual support\u003C\u002Fh3>\n\u003Cp>DeepSeek V4 Pro is trained extensively on Chinese and English, with strong support for other major languages. For Chinese-language applications, DeepSeek is among the best models available — sometimes outperforming Western models on Chinese-language tasks.\u003C\u002Fp>\n\u003Ch3 id=\"mathematical-reasoning\">Mathematical reasoning\u003C\u002Fh3>\n\u003Cp>While not at GPT-5.5's level, DeepSeek V4 Pro is competent at mathematical reasoning. It can solve most high-school and undergraduate-level math problems, and handles arithmetic and algebra reliably. For everyday mathematical tasks, it's more than adequate.\u003C\u002Fp>\n\u003Ch3 id=\"speed\">Speed\u003C\u002Fh3>\n\u003Cp>DeepSeek V4 Pro is fast — its smaller size and optimized architecture mean it generates responses quickly. For applications where latency matters — real-time chat, interactive tools, customer service — DeepSeek V4 Pro delivers responsive performance.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2 id=\"5-where-deepseek-v4-pro-falls-short\">5. Where DeepSeek V4 Pro falls short\u003C\u002Fh2>\n\u003Ch3 id=\"raw-iq-and-complex-reasoning\">Raw IQ and complex reasoning\u003C\u002Fh3>\n\u003Cp>With an IQ of 111, DeepSeek V4 Pro is significantly below the frontier models on complex reasoning tasks. For the hardest problems — competition mathematics, advanced logic puzzles, multi-step abstract reasoning — the frontier models are substantially better. If your task requires genius-level reasoning, DeepSeek V4 Pro will struggle where Claude or GPT-5.5 would succeed.\u003C\u002Fp>\n\u003Ch3 id=\"agentic-tasks\">Agentic tasks\u003C\u002Fh3>\n\u003Cp>On GDPval-AA v2, DeepSeek V4 Pro is in the top 10 but well behind Claude and GPT-5.5. For complex, multi-step real-world tasks — writing large software projects, managing research workflows, executing intricate plans — DeepSeek V4 Pro is less reliable. It can handle simple agentic tasks but struggles with complex ones.\u003C\u002Fp>\n\u003Ch3 id=\"hallucination-rate\">Hallucination rate\u003C\u002Fh3>\n\u003Cp>DeepSeek V4 Pro has a moderate hallucination rate — higher than Claude (35.9%) and Gemini. It's more likely to confidently state incorrect information, particularly on niche or specialized topics. For applications where factual accuracy is critical (research, law, medicine), Claude is more reliable.\u003C\u002Fp>\n\u003Ch3 id=\"emotional-intelligence\">Emotional intelligence\u003C\u002Fh3>\n\u003Cp>DeepSeek V4 Pro's EQ (emotional intelligence) is estimated at ~105 — lower than all frontier models (Claude ~132, GPT-5.5 ~120, Gemini ~115, Grok ~110). Its responses tend to be more mechanical and less nuanced in understanding human emotions. For therapy apps, customer service, or sensitive human interactions, this is a limitation.\u003C\u002Fp>\n\u003Ch3 id=\"safety-and-alignment\">Safety and alignment\u003C\u002Fh3>\n\u003Cp>DeepSeek V4 Pro is subject to Chinese regulatory requirements, which means it may refuse to answer certain politically sensitive questions. This is different from the safety filters of Western models (which focus on harmful content) and may affect its usefulness for certain topics.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2 id=\"6-deepseek-v4-pro-vs-other-models\">6. DeepSeek V4 Pro vs other models\u003C\u002Fh2>\n\u003Cdiv class=\"article-table-wrapper\">\u003Ctable>\n\u003Cthead>\n\u003Ctr>\n\u003Cth>Model\u003C\u002Fth>\n\u003Cth>Intelligence Index\u003C\u002Fth>\n\u003Cth>Mensa Norway IQ\u003C\u002Fth>\n\u003Cth>AI IQ est.\u003C\u002Fth>\n\u003Cth>Hallucination\u003C\u002Fth>\n\u003Cth>Cost\u002F1M\u003C\u002Fth>\n\u003Cth>Open-source\u003C\u002Fth>\n\u003C\u002Ftr>\n\u003C\u002Fthead>\n\u003Ctbody>\u003Ctr>\n\u003Ctd>DeepSeek V4 Pro\u003C\u002Ftd>\n\u003Ctd>44.3\u003C\u002Ftd>\n\u003Ctd>~111\u003C\u002Ftd>\n\u003Ctd>~111\u003C\u002Ftd>\n\u003Ctd>moderate\u003C\u002Ftd>\n\u003Ctd>0.44\u002F0.87\u003C\u002Ftd>\n\u003Ctd>yes\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Claude Opus 4.8\u003C\u002Ftd>\n\u003Ctd>55.7\u003C\u002Ftd>\n\u003Ctd>~130\u003C\u002Ftd>\n\u003Ctd>~132\u003C\u002Ftd>\n\u003Ctd>35.9% (lowest)\u003C\u002Ftd>\n\u003Ctd>5\u002F25\u003C\u002Ftd>\n\u003Ctd>no\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>GPT-5.5\u003C\u002Ftd>\n\u003Ctd>54.8\u003C\u002Ftd>\n\u003Ctd>~145\u003C\u002Ftd>\n\u003Ctd>~136\u003C\u002Ftd>\n\u003Ctd>moderate\u003C\u002Ftd>\n\u003Ctd>5\u002F30\u003C\u002Ftd>\n\u003Ctd>no\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Gemini 3.1 Pro\u003C\u002Ftd>\n\u003Ctd>46.5\u003C\u002Ftd>\n\u003Ctd>141\u003C\u002Ftd>\n\u003Ctd>~131\u003C\u002Ftd>\n\u003Ctd>low\u003C\u002Ftd>\n\u003Ctd>2\u002F12\u003C\u002Ftd>\n\u003Ctd>no\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003Ctr>\n\u003Ctd>Grok-4.20\u003C\u002Ftd>\n\u003Ctd>top 5\u003C\u002Ftd>\n\u003Ctd>145\u003C\u002Ftd>\n\u003Ctd>~140\u003C\u002Ftd>\n\u003Ctd>moderate\u003C\u002Ftd>\n\u003Ctd>3\u002F15\u003C\u002Ftd>\n\u003Ctd>no\u003C\u002Ftd>\n\u003C\u002Ftr>\n\u003C\u002Ftbody>\u003C\u002Ftable>\u003C\u002Fdiv>\n\u003Cp>The picture: DeepSeek V4 Pro is the \u003Cstrong>budget champion\u003C\u002Fstrong> — not the smartest, but by far the most affordable. Claude is the \u003Cstrong>capable professional\u003C\u002Fstrong> — best at real-world tasks and factual accuracy. GPT-5.5 is the \u003Cstrong>test genius\u003C\u002Fstrong> — best at math and visual reasoning. Gemini is the \u003Cstrong>value champion\u003C\u002Fstrong> — best cost-effectiveness among frontier models. Grok-4.20 is the \u003Cstrong>raw IQ champion\u003C\u002Fstrong> — highest IQ with a unique personality.\u003C\u002Fp>\n\u003Cp>For users who need capable AI at the lowest possible price, DeepSeek V4 Pro is the clear choice. It won't match the frontier models on the hardest problems, but for 90% of everyday tasks, it's more than good enough — at a fraction of the cost.\u003C\u002Fp>\n\u003Chr>\n\u003Ch2 id=\"7-what-does-this-mean-for-humans\">7. What does this mean for humans?\u003C\u002Fh2>\n\u003Cp>DeepSeek V4 Pro operates at an IQ of ~111 — smarter than about 25% of humans. But:\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>IQ isn't everything\u003C\u002Fstrong>: an IQ of 111 is above average and sufficient for most practical tasks\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Cost matters more than IQ for most uses\u003C\u002Fstrong>: for everyday applications, the 10x cost savings of DeepSeek often matter more than the 20-35 point IQ gap with frontier models\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Open-source enables innovation\u003C\u002Fstrong>: DeepSeek's open weights allow researchers and developers to build custom solutions, fine-tune for specific domains, and run locally\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Knowledge ≠ understanding\u003C\u002Fstrong>: like all AI models, DeepSeek has access to vast knowledge but doesn't truly "understand" the way a human does\u003C\u002Fli>\n\u003Cli>\u003Cstrong>No consciousness\u003C\u002Fstrong>: high IQ doesn't mean sentience. DeepSeek is a sophisticated pattern-matching system, not a thinking being\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Democratization of AI\u003C\u002Fstrong>: DeepSeek V4 Pro proves that capable AI doesn't have to be expensive — it can be accessible to everyone\u003C\u002Fli>\n\u003C\u002Ful>\n\u003Cp>The most honest answer: DeepSeek V4 Pro is \u003Cstrong>smarter than a quarter of humans on cognitive tests\u003C\u002Fstrong>, and for most practical purposes, it's the smartest choice — not because it's the most intelligent, but because it delivers the most value per dollar.\u003C\u002Fp>\n\u003Cp>\n    \u003Cdiv class=\"article-quiz-cta\">\n      \u003Cdiv class=\"article-quiz-cta__text\">How does your IQ compare to DeepSeek V4 Pro? Take NeuroLab's professional IQ assessment to find out.\u003C\u002Fdiv>\n      \u003Ca href=\"\u002Fen\u002Ftests\u002Fiq-standard\" class=\"article-quiz-cta__button\">\n        Take the Test Now →\n      \u003C\u002Fa>\n    \u003C\u002Fdiv>\u003C\u002Fp>\n\u003Chr>\n\u003Ch2 id=\"conclusion\">Conclusion\u003C\u002Fh2>\n\u003Cp>\n    \u003Cdiv class=\"article-callout\">\n      \u003Cdiv class=\"article-callout__title\">\n        \u003Cspan class=\"article-callout__icon\">💡\u003C\u002Fspan>\n        Key takeaways\n      \u003C\u002Fdiv>\n      \u003Cdiv class=\"article-callout__body\">- \u003Cstrong>DeepSeek V4 Pro's IQ is 111\u003C\u002Fstrong> on the Mensa Norway test — above average human, smarter than ~25% of people.\u003C\u002Fp>\n\u003Cul>\n\u003Cli>\u003Cstrong>44.3 on the Intelligence Index\u003C\u002Fstrong> — top 10 globally, but ~10 points behind frontier models.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Lowest cost of any major model\u003C\u002Fstrong> — 0.44\u002F0.87 per 1M tokens, roughly 10x cheaper than Claude or GPT-5.5.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Open-source\u003C\u002Fstrong> — weights available for local deployment, enabling privacy and customization.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Best for\u003C\u002Fstrong>: cost-sensitive applications, coding, high-volume tasks, Chinese-language applications, local deployment.\u003C\u002Fli>\n\u003Cli>\u003Cstrong>Not best for\u003C\u002Fstrong>: complex reasoning (frontier models win), agentic tasks (Claude wins), factual accuracy (Claude wins), emotional intelligence (Claude wins).\u003C\u002Fli>\n\u003Cli>\u003Cstrong>The lesson\u003C\u002Fstrong>: intelligence isn't just about being the smartest — it's about being accessible. DeepSeek V4 Pro democratizes AI, bringing capable intelligence to everyone.\u003C\u002Fdiv>\n    \u003C\u002Fdiv>\u003C\u002Fli>\n\u003C\u002Ful>\n",10,[18,22,25,28,32,35,38,41,44,47,50,53,56,59,62,65,68,71,74,77,80,83,86,89,92],{"id":19,"text":20,"level":21},"deepseek-v4-pro-china-s-open-source-champion","DeepSeek V4 Pro: China's open-source champion",2,{"id":23,"text":24,"level":21},"1-the-budget-revolution","1. The budget revolution",{"id":26,"text":27,"level":21},"2-benchmark-breakdown","2. Benchmark breakdown",{"id":29,"text":30,"level":31},"mensa-norway-iq-test-111-above-average","Mensa Norway IQ test: 111 (above average)",3,{"id":33,"text":34,"level":31},"artificial-analysis-intelligence-index-v4-1-44-3-top-10","Artificial Analysis Intelligence Index v4.1: 44.3 (top 10)",{"id":36,"text":37,"level":31},"ai-iq-composite-vexowire-111","AI IQ composite (VexoWire): ~111",{"id":39,"text":40,"level":31},"gdpval-aa-v2-top-10","GDPval-AA v2: top 10",{"id":42,"text":43,"level":31},"humanity-s-last-exam-hle-top-10","Humanity's Last Exam (HLE): top 10",{"id":45,"text":46,"level":21},"3-what-does-an-iq-of-111-mean","3. What does an IQ of 111 mean?",{"id":48,"text":49,"level":21},"4-where-deepseek-v4-pro-excels","4. Where DeepSeek V4 Pro excels",{"id":51,"text":52,"level":31},"cost-effectiveness","Cost-effectiveness",{"id":54,"text":55,"level":31},"coding-and-technical-tasks","Coding and technical tasks",{"id":57,"text":58,"level":31},"open-source-availability","Open-source availability",{"id":60,"text":61,"level":31},"multilingual-support","Multilingual support",{"id":63,"text":64,"level":31},"mathematical-reasoning","Mathematical reasoning",{"id":66,"text":67,"level":31},"speed","Speed",{"id":69,"text":70,"level":21},"5-where-deepseek-v4-pro-falls-short","5. Where DeepSeek V4 Pro falls short",{"id":72,"text":73,"level":31},"raw-iq-and-complex-reasoning","Raw IQ and complex reasoning",{"id":75,"text":76,"level":31},"agentic-tasks","Agentic tasks",{"id":78,"text":79,"level":31},"hallucination-rate","Hallucination rate",{"id":81,"text":82,"level":31},"emotional-intelligence","Emotional intelligence",{"id":84,"text":85,"level":31},"safety-and-alignment","Safety and alignment",{"id":87,"text":88,"level":21},"6-deepseek-v4-pro-vs-other-models","6. DeepSeek V4 Pro vs other models",{"id":90,"text":91,"level":21},"7-what-does-this-mean-for-humans","7. What does this mean for humans?",{"id":93,"text":94,"level":21},"conclusion","Conclusion",{"success":4,"data":96,"pagination":265},[97,105,112,120,127,135,142,143,151,158,165,172,179,186,194,201,208,215,222,229,236,243,250,258],{"id":98,"slug":99,"coverImage":100,"author":9,"viewsCount":10,"createdAt":101,"title":102,"description":103,"lang":14,"readingTime":104},"jy0egbpgrn30am622jy7xfrw","adhd-and-iq","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1507812984078-917231447e3e?w=1280","2026-08-01T10:43:58.026Z","ADHD and IQ: When Attention Deficit Hides High Intelligence","Can ADHD mask high IQ? Yes. Research shows that attention deficits can artificially lower IQ test scores by 10-15 points — hiding gifted minds behind poor focus.",7,{"id":106,"slug":107,"coverImage":108,"author":9,"viewsCount":10,"createdAt":109,"title":110,"description":111,"lang":14,"readingTime":104},"dvlwozf321i8jlk9sfl390k1","online-vs-certified-iq-tests","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1499209974431-9fccce68dc04?w=1280","2026-08-01T10:42:47.143Z","Online IQ Tests vs Certified Assessments: How Accurate Are They?","You scored 140 on a free online IQ test. But what does that actually mean? We compare online tests with certified assessments — and the results may surprise you.",{"id":113,"slug":114,"coverImage":115,"author":9,"viewsCount":10,"createdAt":116,"title":117,"description":118,"lang":14,"readingTime":119},"maxnlmh4wqwxjuidbe2cb1jh","iq-above-160","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1454165804606-c3d57bc86b40?w=1280","2026-08-01T10:41:43.616Z","IQ Above 160: What Life Looks Like at the Extreme","Only 1 in 30,000 people have an IQ above 160. We explore what life is like at the extreme edge of human intelligence — the gifts, the struggles, and the myths.",8,{"id":121,"slug":122,"coverImage":123,"author":9,"viewsCount":10,"createdAt":124,"title":125,"description":126,"lang":14,"readingTime":16},"z8wcdr5vus4cvw8zr3p1ok1r","iq-llama-4","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1526374965328-4f18af581d06?w=1280","2026-08-01T10:00:01.170Z","What is the IQ of Llama 4?","Llama 4 scores 112 on the Mensa Norway IQ test and ranks in the top 15 on the Intelligence Index. We break down every benchmark and explain what makes Meta's open-source model unique.",{"id":128,"slug":129,"coverImage":130,"author":9,"viewsCount":131,"createdAt":132,"title":133,"description":134,"lang":14,"readingTime":16},"pxehdebhlpxfs0h5qm2m23e9","iq-qwen-35","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1518186285589-2f7649de83e0?w=1280",1,"2026-08-01T09:38:28.903Z","What is the IQ of Qwen 3.5?","Qwen 3.5 scores 115 on the Mensa Norway IQ test and ranks in the top 10 on the Intelligence Index. We break down every benchmark and explain what makes Alibaba's model unique.",{"id":136,"slug":137,"coverImage":138,"author":9,"viewsCount":21,"createdAt":139,"title":140,"description":141,"lang":14,"readingTime":16},"st1nxd9ef5blcslk68p83pj1","iq-kimi-k3","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1620712943524-bccbe725b837?w=1280","2026-08-01T09:29:37.632Z","What is the IQ of Kimi K3?","Kimi K3 scores 118 on the Mensa Norway IQ test and ranks in the top 10 on the Intelligence Index. We break down every benchmark and explain what makes Moonshot AI's model unique.",{"id":6,"slug":7,"coverImage":8,"author":9,"viewsCount":10,"createdAt":11,"title":12,"description":13,"lang":14,"readingTime":16},{"id":144,"slug":145,"coverImage":146,"author":9,"viewsCount":147,"createdAt":148,"title":149,"description":150,"lang":14,"readingTime":16},"lci1u8drubje01qwq1vah01m","iq-grok-420","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1635776062127-d379b0ba009b?w=1280",5,"2026-07-31T20:28:41.558Z","What is the IQ of Grok-4.20?","Grok-4.20 scores 145 on the Mensa Norway IQ test — tied for #1 among all AI models. We break down every benchmark and explain what makes xAI's model unique.",{"id":152,"slug":153,"coverImage":154,"author":9,"viewsCount":21,"createdAt":155,"title":156,"description":157,"lang":14,"readingTime":16},"n32qvrqd1er095jgz98qg73l","iq-gemini-31-pro","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1591488320449-011701bb6704?w=1280","2026-07-31T20:04:57.324Z","What is the IQ of Gemini 3.1 Pro?","Gemini 3.1 Pro scores 141 on the Mensa Norway IQ test and 46.5 on the Intelligence Index. We break down every benchmark and explain why it's the most cost-effective frontier model.",{"id":159,"slug":160,"coverImage":161,"author":9,"viewsCount":131,"createdAt":162,"title":163,"description":164,"lang":14,"readingTime":16},"jl69290gswdg1wrb5oy6iiyo","iq-gpt-55","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1620712943543-bcc4688e7485?w=1280","2026-07-31T19:57:55.350Z","What is the IQ of GPT-5.5?","GPT-5.5 scores ~145 on the Mensa Norway IQ test and 54.8 on the Intelligence Index. We break down every benchmark, explain what the numbers mean, and compare it to Claude, Gemini, and Grok.",{"id":166,"slug":167,"coverImage":168,"author":9,"viewsCount":131,"createdAt":169,"title":170,"description":171,"lang":14,"readingTime":16},"fmyy2tdkvdmffu6n9y5npc5c","iq-claude-opus-48","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1677442136019-21780ecad995?w=1280","2026-07-31T19:51:43.806Z","What is the IQ of Claude Opus 4.8?","Claude Opus 4.8 tops the Artificial Analysis Intelligence Index at 55.7 and scores ~132-145 on human IQ tests. We break down every benchmark, explain what the numbers mean, and compare it to GPT-5.5, Gemini, and Grok.",{"id":173,"slug":174,"coverImage":175,"author":9,"viewsCount":10,"createdAt":176,"title":177,"description":178,"lang":14,"readingTime":119},"mhlz1hmh5ez4vs6loftlxd06","fasting-cognitive-performance","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1490645935441-5b6c2cc1e5e1?w=1280","2026-07-30T18:25:44.150Z","Does Fasting Improve Cognitive Performance?","From intermittent fasting to ketones: we review the evidence on whether fasting sharpens your mind, boosts focus, or actually impairs cognition. The science may surprise you.",{"id":180,"slug":181,"coverImage":182,"author":9,"viewsCount":21,"createdAt":183,"title":184,"description":185,"lang":14,"readingTime":16},"fentmgsgc3lq034m2nk53ol4","wechsler-iq-test-subtests","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1494256997604-668ec142f5d2?w=1280","2026-07-30T18:15:36.675Z","Wechsler IQ Test: What Each Subtest Really Measures","The WAIS and WISC are the most widely used IQ tests in the world. We break down every subtest, what it actually measures, and how scores translate to real cognitive abilities.",{"id":187,"slug":188,"coverImage":189,"author":9,"viewsCount":10,"createdAt":190,"title":191,"description":192,"lang":14,"readingTime":193},"wlwyq33ezei7c22edclwpyvo","lead-fluoride-iq-loss","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1581093577124-a6f4b6e3a5e1?w=1280","2026-07-30T18:04:10.914Z","Lead, Fluoride, and IQ Loss: The Environmental Evidence","Can environmental toxins actually lower your IQ? We review the strongest studies on lead exposure, fluoride, and cognitive decline — separating solid science from sensationalism.",9,{"id":195,"slug":196,"coverImage":197,"author":9,"viewsCount":31,"createdAt":198,"title":199,"description":200,"lang":14,"readingTime":119},"ue16gxaun3lus9vml8imf6ci","iq-changes-during-pregnancy","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1559750178-961c031777fe?w=1280","2026-07-29T18:43:46.337Z","IQ Changes During Pregnancy: What Research Finds","Does pregnancy really shrink your brain? We review the neuroscience of maternal cognition, from \"pregnancy brain\" to lasting structural changes, and separate fact from myth.",{"id":202,"slug":203,"coverImage":204,"author":9,"viewsCount":131,"createdAt":205,"title":206,"description":207,"lang":14,"readingTime":193},"zd0sfdazkyai41s0kh639ohl","does-bilingualism-raise-iq","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1523050854058-8f908275853f?w=1280","2026-07-29T18:29:58.855Z","Does Speaking Two Languages Actually Raise Your IQ?","The bilingual advantage has been debated for decades. We review the strongest studies on bilingualism and cognitive ability, separating real effects from hype.",{"id":209,"slug":210,"coverImage":211,"author":9,"viewsCount":31,"createdAt":212,"title":213,"description":214,"lang":14,"readingTime":193},"s0hko69oudyziplx2q04ab9q","gifted-kids-peaked-early-childhood-iq-fades","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1503454537195-1dc81782b6d3?w=1280","2026-07-29T18:09:49.555Z","Gifted Kids Who Peaked Early: Why Childhood IQ Fades","Why do so many child prodigies become ordinary adults? We examine the research on early IQ peaking, the \"regression to the mean\" effect, and what actually predicts adult success.",{"id":216,"slug":217,"coverImage":218,"author":9,"viewsCount":147,"createdAt":219,"title":220,"description":221,"lang":14,"readingTime":119},"s7x75o02kixj2u43gw4ah5pa","olfactory-processing-speed-cognitive-marker","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1542838132-92c53300287c?w=1280","2026-07-28T20:38:15.776Z","Olfactory Processing Speed as a Cognitive Marker","Can how fast you identify smells predict your cognitive ability? We examine the surprising research linking olfactory processing speed to IQ, memory, and neurodegenerative disease risk.",{"id":223,"slug":224,"coverImage":225,"author":9,"viewsCount":21,"createdAt":226,"title":227,"description":228,"lang":14,"readingTime":193},"v1qq1yz082k72zzynk0qv2yh","do-musicians-have-higher-iqs","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1520006510097-3d64427f3f98?w=1280","2026-07-28T20:25:14.220Z","Do Musicians Have Higher IQs? The Research Verdict","The idea that musicians are smarter is widespread — but does the science support it? We review 10 key studies on musical training, IQ, and cognitive ability to separate fact from fiction.",{"id":230,"slug":231,"coverImage":232,"author":9,"viewsCount":31,"createdAt":233,"title":234,"description":235,"lang":14,"readingTime":193},"xam91ejelkjg23x2z9skt9vm","are-night-owls-smarter","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1518837695005-2083093ee35b?w=1280","2026-07-28T19:47:32.529Z","Are Night Owls Smarter? Chronotype and Cognitive Performance","The idea that night owls are smarter than early birds is popular — but what does the research actually show? We examine the evidence on chronotype, IQ, and cognitive performance.",{"id":237,"slug":238,"coverImage":239,"author":9,"viewsCount":147,"createdAt":240,"title":241,"description":242,"lang":14,"readingTime":193},"hpakkt3uv7frkkhj2el6fvm1","does-high-iq-make-you-rich","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1579621970795-87facc2f976d?w=1280","2026-07-28T19:37:46.682Z","Does High IQ Make You Rich? 12 Studies Reviewed","The correlation between IQ and income is real but surprisingly weak. We review 12 key studies to understand why high IQ doesn't guarantee wealth — and what actually predicts financial success.",{"id":244,"slug":245,"coverImage":246,"author":9,"viewsCount":147,"createdAt":247,"title":248,"description":249,"lang":14,"readingTime":193},"b5ct19s9mfso606wcppcp79o","iq-test-scores-in-prison-populations","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1497366216548-37526070297c?w=1280","2026-07-28T18:14:35.540Z","IQ Test Scores in Prison Populations: What the Studies Show","Research consistently finds that prison populations score lower on IQ tests than the general public — by about 10-12 points on average. But the reasons are far more complex than \"criminals are less intelligent.\" This article examines the evidence, the confounding factors, and what the data actually means.",{"id":251,"slug":252,"coverImage":253,"author":9,"viewsCount":254,"createdAt":255,"title":256,"description":257,"lang":14,"readingTime":16},"zjjd6y8yfcydy6y02c9xa2yf","which-country-has-the-highest-average-iq","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1526129313847-8d262a6b1f0d?w=1280",6,"2026-07-28T17:59:21.857Z","Which Country Has the Highest Average IQ (And Why the Data Is Messy)","Every few months, a new headline declares that Singapore, Hong Kong, or South Korea has the highest IQ in the world. But behind these rankings lies a tangle of methodological problems, cultural biases, and outdated data. This article unpacks what the research actually shows — and what it doesn't.",{"id":259,"slug":260,"coverImage":261,"author":9,"viewsCount":254,"createdAt":262,"title":263,"description":264,"lang":14,"readingTime":193},"nxfpaqozmm0dpaytyy2cjixv","why-your-iq-test-score-changes-over-time","https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1499750310107-5fef28a66643?w=1280","2026-07-28T17:33:41.377Z","Why Your IQ Test Score Changes Over Time","Your IQ score is not a fixed number stamped on your brain. Research shows that IQ scores can change significantly over a lifetime — sometimes by 20 points or more. This guide explains what drives these changes and what they mean for you.",{"page":131,"pageSize":266,"pageCount":31,"total":267},24,49]