France's Mistral unveils its first 'reasoning' AI model
French artificial intelligence startup Mistral on Tuesday announced a so-called "reasoning" model it said was capable of working through complex problems, following in the footsteps of top US developers.
Available immediately on the company's platforms as well as the AI platform Hugging Face, the Magistral "is designed to think things through - in ways familiar to us," Mistral said in a blog post.
The AI was designed for "general purpose use requiring longer thought processing and better accuracy" than its previous generations of large language models (LLMs), the company added.
Like other "reasoning" models, Magistral displays a so-called "chain of thought" that purports to show how the system is approaching a problem given to it in natural language.
This means users in fields like law, finance, healthcare and government would receive "traceable reasoning that meets compliance requirements" as "every conclusion can be traced back through its logical steps", Mistral said.
The company's claim gestures towards the challenge of so-called 'interpretability,' or working out how AI systems arrive at a given response.
Since they are "trained" on gigantic corpuses of data rather than directly programmed by humans, much behaviour by AI systems remains impenetrable even to their creators.
Mistral also vaunted improved performance in software coding and creative writing by Magistral.
Competing "reasoning" models include OpenAI's o3, some versions of Google's Gemini and Anthropic's Claude, or Chinese challenger DeepSeek's R1.
The idea that AIs can "reason" was called into question this week by Apple, the tech giant that has struggled to match achievements by leaders in the field.
Several Apple researchers published a paper called "The Illusion of Thinking" that claimed to find "fundamental limitations in current models" which "fail to develop generalisable reasoning capabilities beyond certain complexity thresholds".
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Indian Express
23 minutes ago
- Indian Express
In a first, UN report calls for urgent action on AGI: What does it mean?
The United Nations Council of Presidents of the General Assembly (UNCPGA) recently released a report seeking immediate global coordination in the wake of the emergence of artificial general intelligence (AGI). At present, leading tech giants OpenAI, Google, and Meta are working towards achieving AGI, which are essentially systems that are capable of equalling or surpassing human intelligence in various cognitive tasks. OpenAI's Sam Altman has been saying AGI may be on the horizon although these human intelligence-level systems aren't here yet. Similarly, Google DeepMind, as a step towards AGI, is reportedly working towards creating a 'world-modelling' team to simulate environments. Meta, on the other hand, is reportedly investing $15 billion into Scale AI and a 50-member squad to accelerate its AGI vision. Anthropic is working on safe and adaptable models, and there are speculations that it will reach AGI in the next two to three years. So far, no company has demonstrated true AGI, and most of these are scaling up multimodal and reinforcement-learning systems. Even though AGI seems to be a work in progress, the UNCPGA report suggests that with massive financial investments in history and unprecedented R&D efforts, AGI could emerge within this decade. It could lead to some extraordinary benefits, such as the acceleration of scientific discoveries related to public health, transformation of industries and increased productivity, and even contribute to the realisation of sustainable development goals. The report also acknowledges that AGI could lead to unique and potentially 'catastrophic risks'. AGI could execute harmful actions beyond human oversight, resulting in irreversible impacts. 'Without proactive global management, competition among nations and corporations will accelerate risky AGI development, undermine security protocols, and exacerbate geopolitical tensions. Coordinated international action can prevent these outcomes, promoting secure AGI development and usage, equitable distribution of benefits, and global stability,' reads the report. The UNCPGA report recommends immediate and coordinated international action supported by the United Nations to effectively address the global challenges that could arise owing to AGI. The report underscores six major risks, including loss of human control over AGI; AGI-enabled weapons of mass destruction; cybersecurity vulnerabilities; economic instability; autonomous AGI with existential risks; and missed opportunities to solve global challenges. On the other hand, the UN panel also listed its recommendations to mitigate these risks, which include bold and coordinated steps, including a dedicated UN General Assembly session on AGI; a global AGI Observatory to track the development and risks; a certification system for secure and trustworthy AGI; A proposed UN framework convention on AGI governance; and exploring a dedicated UN agency for global coordination.
India Today
34 minutes ago
- India Today
iPhone 15 price cut by Rs 10,600
iPhone 15 price cut by Rs 10,600 By Ankita Garg Amazon is not running any specific sale, but is offering a big discount of Rs 10,600 on the iPhone 15, suggesting that the device has received a price cut on the platform. Here are the details. The iPhone 15 is listed on Amazon at Rs 59,300, down from its current official retail price of Rs 69,900. So, you are getting a flat discount of Rs 10,600. The deal is for black and pink colours. This is a great deal for people who are looking to upgrade from older iPhones. The mentioned price is for the 128GB storage model. The iPhone 15 is one of the best iPhones from Apple and it has a compact 6.1-inch display and comes in five vibrant colours: pink, yellow, green, blue, and black. The device doesn't offer support for Apple Intelligence (AI), but it will still offer a good experience to power users. Apple is anyway yet to make its AI game strong, so iPhone 15 buyers won't be missing out on anything. With a 48-megapixel primary camera, the iPhone 15 captures stunning photos in various lighting conditions. Apple says the battery of iPhone 15 is designed for all-day usage, but people can expect around 9 hours of battery on an average. The A16 Bionic chip provides improved speed and efficiency compared to the previous generation. The iPhone 15 also adopts the universal USB Type-C charging port, replacing the Lightning connector.
Indian Express
an hour ago
- Indian Express
Who is Alexandr Wang, and why is Meta betting billions on his startup Scale AI?
Alexandr Wang is the CEO and co-founder of Scale AI, a data-labelling startup that helps other companies train and deploy cutting-edge AI models. Over the years, Wang has built his startup into the backbone of the AI boom, quietly enabling everything from autonomous vehicles to large language models (LLMs). Now, Wang finds himself at the centre of a potential $15 billion shake-up as Meta taps him to lead its newly formed research lab that will focus on building AI systems capable of 'superintelligence'. The $15 billion investment deal is also expected to bring other Scale AI employees to Meta, which is also reportedly offering seven to nine-figure compensation packages to AI researchers from OpenAI and Google who would like to be a part of its new 50-member artificial superintelligence lab. The new lab comes at a crucial time for Meta, which is perceived to be struggling to pull ahead of its competitors Google, Microsoft, and OpenAI in the high-stakes AI race. CEO Mark Zuckerberg has pushed for AI to be incorporated across the company's products such as its Ray Ban smart glasses as well as social media platforms Facebook, Instagram, and WhatsApp. Meta has also sought to define its competitive edge by developing open AI models, allowing developers to freely download and integrate the source code into their own tools. But internal issues such as employee turnover and underwhelming product launches have reportedly hampered Meta's AI efforts lately. So far, the company's research efforts have been overseen by its chief AI scientist, Turing Award winner Yann LeCun who is widely recognised for his groundbreaking research contributions on convolutional neural networks (CNNs). However, LeCun's views on AI are not aligned with others in Silicon Valley as he has argued that LLMs are not the path to artificial general intelligence (AGI). Now, Meta is betting on Wang to not only help it regain the lead in the AI race but also push toward another frontier known as artificial superintelligence (ASI) — a hypothetical AI system with intelligence exceeding that of the human brain. Alexandr Wang was born in New Mexico, US, to Chinese immigrant parents who worked at Los Alamos National Laboratory as nuclear physicists. Before heading to college, Wang reportedly worked at internet startup Quora. He dropped out of Massachusetts Institute of Technology (MIT) after just one year and joined Y Combinator, the popular startup accelerator that used to be led by OpenAI CEO Sam Altman. At Y Combinator, he teamed up with Quora alum Lucy Guo to start a new company called Scale AI in 2016. Two years later, both Wang and Guo were named in Forbes' 30 Under 30 list in enterprise technology. Guo shortly exited Scale AI 'due to differences in product vision and road map,' according to a report by Forbes. Wang continued running the startup which was minted as a unicorn in 2019 after raising $100 million in investment from Peter Thiel's Founders Fund followed by another $580 million fundraising round which put the company at a $7 billion valuation. At 24, Wang became the youngest self-made billionaire in the world. His co-founder, Lucy Guo, recently became the youngest self-made woman billionaire due to her stake in Scale AI. Wang was reportedly Sam Altman's roommate during the COVID-19 pandemic. The two AI industry leaders were also photographed sitting next to each other at US President Donald Trump's swearing-in ceremony in January this year. Scale AI was founded in 2016 as a startup that labelled mass quantities of data required to train AI systems, particularly autonomous vehicles (AV). As a result, most of its data services were primarily offered to self-driving automakers. This move to corner the market for supplying training data so that self-driving cars could tell the difference between various objects, is what made Scale AI well-positioned in the AI boom that was soon to follow. LLMs are trained on massive amounts of data to generate text and other content. Scale AI hires thousands of contract workers to sift through vast amounts of data, label the information, and clean the datasets that are then supplied to tech companies to train their complex AI models. Scale AI's client list includes major automakers such as Toyota and Honda as well as Waymo, Google's AV subsidiary. It has also partnered with Accenture to help the consulting giant build custom AI apps and models. OpenAI, Microsoft, and Toronto-based AI startup Cohere also count among Scale AI's customers, according to a report by Forbes. The US government has also reportedly sought Scale AI's data labelling and annotation services in order to help analyse satellite imagery in Ukraine. Last valued at nearly $14 billion, the company saw about $870 million in revenue in 2024. It further expects to more than double revenue this year to $2 billion, which would put Scale AI's valuation at $25 billion, according to a report by Bloomberg. However, the AI boom has also given rise to a wave of relatively new competitors such as Surge AI, which offers data labeling tools to AI companies, as well as data labeling startups Labelbox and Snorkel AI, which primarily cater to non-tech enterprises.
