Vlad Tenev, the CEO of Robinhood, co-founded the AI (Artificial Intelligence) business Harmonic, based in New York, which on Monday announced the beta release of an iOS and Android chatbot app that allows customers to access its AI model, Aristotle. Currently in beta testing for iOS and Android, the app is called Aristotle after the AI model that drives it. The launch occurs weeks after the business made the iOS app’s waitlist available for interested parties to sign up for. With the release of the apps, Harmonic claims to also make its Aristotle AI model freely available, allowing users to test out its sophisticated mathematical capabilities.
With this launch, Harmonic hopes to increase access to Aristotle, which it says provides “hallucination-free” solutions to mathematical reasoning questions. This is a bold claim considering the reliability issues with current AI models. The goal of Harmonic is to develop “mathematical superintelligence,” or MSI; in the long run, the business hopes to assist customers in all math-dependent domains, such as computer science, physics, and statistics.
Both the iOS and Android apps are now available in beta, the company said in a news release (via Business Wire). Like Gemini, ChatGPT, and Grok, it is a chatbot app that lets users type questions in natural language and get answers. According to Harmonic, the app’s AI model was designed to be devoid of hallucinations.
In the context of artificial intelligence, hallucinations are situations in which a chatbot confidently gives a false or deceptive response. These answers may spread false information and cast doubt on the accuracy and dependability of AI algorithms. It is known that hallucinations occur to some extent in all large language models (LLMs). Harmonic, however, asserts otherwise.
Additionally, the AI startup reported that Aristotle performed at the level of a gold medal at the 2025 International Mathematical Olympiad (IMO). Both Google and OpenAI previously declared that they had achieved comparable results in the competition.
Tudor Achim, Harmonic’s CEO and co-founder, stated, “The algorithmic innovations that yielded not just IMO Gold-Medal performance, but formally verified solutions to complex, novel mathematical problems, solidify Harmonic’s position at the forefront of AI model development.” He further said that it is the first product available to people that does reasoning and formally verifies the output.” “We actually do guarantee that there are no hallucinations within the domains that Aristotle supports, which are domains of quantitative reasoning.”
In the future, Harmonic says it also intends to release a web application for consumers and an API that would allow businesses to utilize Aristotle.
The company says that Aristotle verifies the answers using its mathematical superintelligence (MSI) idea. It is claimed that the method keeps hallucinations at bay by guaranteeing cogent logical reasoning and highlighting mistakes and contradictions. None of these allegations could be confirmed by Gadgets 360 employees. Additionally, the business has not yet disclosed any benchmark scores for the model.
Harmonic stated that in addition to releasing the AI chatbot as an iOS and Android mobile app, the model will also be accessible online and through an application programming interface (API).
Aristotle won a gold medal at the 2025 International Math Olympiad (IMO), according to Harmonic, after completing a rigorous test in which the problems were converted into a machine-readable format. Through informal tests conducted in natural language, Google and OpenAI also created AI models that performed gold on this year’s IMO.
Claims by Harmonic says it will not be providing any more Aristotelian benchmarks at this moment.
Only a few weeks have passed since Harmonic secured $100 million at a $875 million valuation in a Series B financing headed by Kleiner Perkins, which coincided with the beta launch of Aristotle. According to Achim, investors thought that was a fair investment considering the size of his startup’s ambitions, and Harmonic is “advancing very rapidly along” its road to reaching MSI.
The goal of several top Information tech companies is to train their AI models to tackle mathematical problems. While math is regarded as a uniquely verifiable area that requires fundamental reasoning abilities, AI that can perform math is valuable in and of itself. Systems that acquire these skills might also be beneficial in other fields.
Achim also claims that by using the open source programming language Lean to generate responses from Aristotle, Harmonic is able to offer hyper-accurate solutions. Aristotle claims that the model uses an algorithmic procedure independent of artificial intelligence to confirm that the answer is accurate before providing it to users. Similar technology is used to validate outputs in high-stakes industries including aviation and medical devices, according to the CEO of Harmonic.
It is extremely challenging to get an AI model to work without hallucinations, even in a limited area. Even top AI models frequently experience hallucinations, according to studies, and the issue doesn’t seem to be improving. The more recent AI reasoning models from OpenAI exhibit more hallucinations than their predecessors.
We use cookies to ensure that we give you the best experience on our website. If you continue to use this site we will assume that you are happy with it.
