- Tests from OmniCalculator suggest that Claude and ChatGPT are not the smartest
- The report finds that Grok 4.2 performs best in logic and problem solving
- Claude still leads in writing quality and tone
ChatGPT is still the most popular AI chatbot, even with the exodus coming to Claude, but is it the smartest? A new report from OmniCalculator suggests that ChatGPT might not be the smartest AI out there.
When it comes to the quantifiable mathematical ability of these AI chatbots, the smartest free AI model is, rather surprisingly, Grok. xAI’s Grok 4.2 model specifically. That doesn’t mean anything about its writing style and abilities, or anything else chatbots can do, but it does suggest that it might have the edge in math prowess.
Claude’s winning style
Claude’s recent rise in popularity has been driven by people wanting to leave ChatGPT because of unpopular AI military deals, but also by how it composes replies and writes its replies.
The article continues below
The quality is difficult to quantify in relation to mathematical skills, but easy to recognize. The OmniCalculator report singled out Claude 4.6 as the best at it, able to process and respond to long documents without losing coherence and maintaining a consistent voice throughout. For the average person, this is much more important than which AI can navigate complex logic and math problems.
It even comes out in the facsimiles of personality that the AI models offer. Claude is more willing to acknowledge uncertainty, which can make its responses feel measured rather than overconfident. That tone can create the impression of deeper thinking, regardless of underlying reasoning.
Older models, including earlier versions of ChatGPT and Claude, were found to revise or guess their own answers about 60% of the time in complex problem-solving scenarios. That kind of instability doesn’t always show up in casual use, but it becomes apparent when you push these systems through multi-step reasoning tasks where consistency is important.
But the Grok 4.2 lowers that instability rate down to 33.1%, meaning it’s far less likely to backtrack or change its conclusions mid-process. It’s great for reasoning and logic, but not much help in emulating the smooth tones that make other models feel more polished.
Special subject
The difference in ability is not trivial. Good writing and strong reasoning skills (or the AI facsimiles thereof) are related skills, but they are not identical. A model can produce elegant prose while making subtle errors in logic. Another may arrive at the correct answer but speak in clumsy ways that seem very outdated.
However, the margins are narrow and no model works flawlessly. Even the best make mistakes, sometimes on relatively simple problems. The idea of a single smartest AI is a bit nonsensical in that way. The clear winner in one context may fall back in another.
And there is no such thing as a permanent winner. Each of the leading models fills up slightly differently. Similarly, the underlying complexity of what people mean by intelligence is complex and constantly evolving. Which AI chatbot to trust is situational. The best model for drafting an email may not be the best for solving a technical problem. The most reliable coding assistant may not produce the most natural-sounding text.
As competition intensifies, companies are likely to lean further into their strengths and refine specific capabilities rather than chasing an all-in-one solution. The result could be a landscape where specialization matters as much as scale. So the question of which AI is smarter will probably always have the answer, “depends.”
Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews and opinions in your feeds. Be sure to click the Follow button!
And of course you can too follow TechRadar on TikTok for news, reviews, video unboxings, and get regular updates from us on WhatsApp also.
The best business laptops for all budgets
