OpenAI Unveils O1 Series: A New Frontier in AI with Advanced Reasoning Capabilities

OpenAI o1 and o1-mini models for advanced STEM reasoning unveiled

OpenAI on Thursday announced the launch of OpenAI o1, a new series of large language models designed to enhance reasoning capabilities through reinforcement learning. The model, part of a preview release, focuses on improving complex reasoning by producing a "chain of thought" before responding. This approach aims to make the model more capable in areas such as science, math, and coding compared to previous models, according to the company. OpenAI's latest model series, OpenAI o1, enhances complex problem-solving by using reinforcement learning to refine responses and correct errors. In tests, OpenAI o1 matches the performance of human PhD students in physics, chemistry, and biology and excels in mathematics and programming. While it advances AI capabilities for reasoning tasks, it does not include features like web browsing or file and image uploads found in earlier models. OpenAI continues to refine the model to make it as user-friendly as previous versions while maintaining high performance in reasoning tasks. OpenAI has improved model safety and alignment by incorporating chain of thought reasoning, embedding safety rules directly into the model's reasoning process. The o1-preview model has achieved a score of 84 out of 100 on challenging jailbreak tests, significantly higher than GPT-4o's 22. This enhancement is supported by comprehensive testing, red-teaming, and collaboration with U.S. and U.K. AI Safety Institutes, with results detailed in the System Card. While OpenAI has developed a method to monitor the model's reasoning internally, they have opted not to display the raw chain of thought to users. Instead, users will see a model-generated summary of the chain of thought. This approach balances the need for transparency with user experience and competitive considerations. The OpenAI o1 models are designed for complex problem-solving in areas such as science, coding, and math. Potential applications include annotating cell sequencing data, generating mathematical formulas for quantum optics, and executing multi-step workflows in development. Alongside OpenAI o1, the company also unveiled o1-mini, a cost-efficient reasoning model designed to excel in STEM fields. While it closely matches the performance of OpenAI o1 on benchmarks such as AIME and Codeforces, o1-mini aims to offer a faster and more affordable option for applications requiring focused reasoning without extensive world knowledge. Large language models like o1 are pre-trained on extensive text datasets, providing broad world knowledge but often being costly and slow for specific applications. In contrast, o1-mini is optimized for STEM reasoning, having undergone training with a high-compute reinforcement learning (RL) pipeline similar to o1. This optimization allows o1-mini to perform effectively on various reasoning tasks while being more cost-efficient. Human raters found o1-mini preferable to GPT-4o in reasoning-heavy domains, but less favored in language-focused areas. For tasks requiring complex reasoning, o1-mini often reaches conclusions 3-5 times faster than GPT-4o. o1-mini uses the same alignment and safety methods as o1-preview, demonstrating 59% greater jailbreak resilience on the StrongREJECT dataset compared to GPT-4o. OpenAI has assessed the safety of o1-mini with rigorous evaluations and will release detailed results in the system card. While o1-mini excels in STEM reasoning, its factual knowledge in non-STEM areas is limited compared to larger models like GPT-4o. OpenAI plans to address these limitations in future updates and explore extending the model to other domains. This early preview of the o1 models is available in ChatGPT and the API. Future updates will include browsing, file and image uploading, and further model developments.

OpenAI Unveils 'O1' Model: A Step Closer To Human-Like AI, But Not Without Flaws

Balancing affordability with the power of its bigger counterpart, 'o1-mini' offers a glimpse into the future of AI. OpenAI, the AI research lab, has launched a new model, 'o1', which is the first in a series of "reasoning" models. This model can answer complex questions faster than a human. The company has also introduced a smaller, more affordable version, 'o1-mini'. What Happened: The o1 model is a significant step towards OpenAI's goal of achieving human-like artificial intelligence. It outperforms previous models in writing code and solving multi-step problems. However, it is more expensive and slower to use than the GPT-4o model, reported The Verge. OpenAI is calling the release of o1 a "preview" to highlight its early stage. ChatGPT Plus and Team users can access both o1-preview and o1-mini starting Thursday, while Enterprise and Edu users will get access early next week. The company plans to offer o1-mini access to all free ChatGPT users but has not set a release date yet. Despite its advanced capabilities, o1 has limitations. It is not as proficient as GPT-4o in areas such as factual knowledge about the world, browsing the web, or processing files and images, The Verge reports. However, OpenAI believes it represents a new class of capabilities and has named it o1 to indicate "resetting the counter back to 1." See Also: Elon Musk Says X Intends To Obey The Law In Each Jurisdiction, But If Free Speech Is Not Illegal Then 'What Are We Doing' OpenAI's research lead, Jerry Tworek, explained that the training behind o1 is fundamentally different from its predecessors. The model has been trained to solve problems on its own using reinforcement learning, a technique that teaches the system through rewards and penalties. This new training methodology is expected to make the model more accurate. Despite its limitations, the o1 model has shown remarkable performance in certain areas. The Verge reports that it has outperformed GPT-4o in solving complex problems, such as coding and math, and has been able to explain its reasoning, according to OpenAI. OpenAI Attracts Major Investors: Meanwhile, the United Arab Emirates' state-supported firm, MGX, is reportedly in talks to invest in OpenAI, which could potentially be part of a multibillion-dollar funding round, The Wall Street Journal reported on Thursday. Established by the UAE earlier this year, MGX is primarily focused on investing in artificial intelligence projects. The exact amount of the potential investment in OpenAI is yet to be disclosed. However, if the investment is finalized, it would bring the Middle Eastern nation closer to one of the world's leading AI companies, according to WSJ. OpenAI is reportedly in discussions to raise $6.5 billion in equity financing, potentially bringing its valuation to $150 billion. This valuation is expected to be led by Thrive Capital and does not factor in the new funds being raised. OpenAI's CEO, Sam Altman, has been seeking the U.S. government's support for a project that aims to form a global coalition of investors to fund the costly physical infrastructure required for rapid AI development. This AI infrastructure project is estimated to cost "tens of billions of dollars." Read Next: Bank Of America Warns Energy Could Be 'Cheap For A Reason,' Highlights Utilities As Defensive Play This content was partially produced with the help of AI tools and was reviewed and published by Benzinga editors. Photo: rafapress/Shutterstock.com Market News and Data brought to you by Benzinga APIs

OpenAI releases new o1 models with reasoning capabilities; available to these users - Times of India

Sam Altman-led OpenAI has launched a new series of reasoning models under the o1 series. The company says that these new models can reason through complex tasks and solve harder problems than previous models in science, coding, and math. OpenAI o1 and OpenAI o1-mini, the company said, are trained to spend more time thinking through problems before they respond, much like a person would. OpenAI further claims that the o1 series models perform "similarly to PhD students on challenging benchmark tasks in physics, chemistry, and biology. We also found that it excels in math and coding." "In a qualifying exam for the International Mathematics Olympiad (IMO), GPT-4o correctly solved only 13% of problems, while the reasoning model scored 83%," it added. OpenAI says that the o1 model can be used by healthcare researchers to annotate cell sequencing data, by physicists to generate complicated mathematical formulas needed for quantum optics, and by developers in all fields to build and execute multi-step workflows. OpenAI o1-mini The company has also announced OpenAI o1-mini - a faster, cheaper reasoning model, particularly for coding. "As a smaller model, o1-mini is 80% cheaper than o1-preview, making it a powerful, cost-effective model for applications that require reasoning but not broad world knowledge," the company stated. Availability of OpenAI o1 series The company says that OpenAI o1 series will be available for ChatGPT Plus and Team users. They are rolling out to end users. Both o1-preview and o1-mini can be selected manually in the model picker. The models have a weekly rate limit of 30 messages for o1-preview and 50 for o1-mini. "We are working to increase those rates and enable ChatGPT to automatically choose the right model for a given prompt," the company added. The TOI Tech Desk is a dedicated team of journalists committed to delivering the latest and most relevant news from the world of technology to readers of The Times of India. TOI Tech Desk's news coverage spans a wide spectrum across gadget launches, gadget reviews, trends, in-depth analysis, exclusive reports and breaking stories that impact technology and the digital universe. Be it how-tos or the latest happenings in AI, cybersecurity, personal gadgets, platforms like WhatsApp, Instagram, Facebook and more; TOI Tech Desk brings the news with accuracy and authenticity.

Sam Altman Highlights OpenAI o1 Capabilities In New Launch

OpenAI introduced o1-mini, an affordable version, as part of its new AI lineup, focusing on advanced coding without broad world knowledge. Sam Altman's OpenAI has launched its much-anticipated o1-preview series of AI models. This marks a significant milestone in artificial intelligence development, especially for ChatGPT. These new models excel in solving complex problems in areas such as science, coding, and mathematics. The new models are now available in ChatGPT and the API as part of an early preview, with plans for regular updates and improvements. OpenAI CEO Sam Altman expressed his excitement for the release. He stated on X (formerly Twitter), "extremely proud of the team; this was a monumental effort across the entire company. hope you enjoy it!" The o1-preview models introduce a new level of reasoning, spending more time processing information before generating responses. This refinement leads to improved problem-solving abilities. In initial tests, the next update of the reasoning model performed on par with PhD students in physics, chemistry, and biology tasks, and achieved impressive results in math and coding competitions. For example, in a qualifying exam for the International Mathematics Olympiad, the o1 model scored an impressive 83%, compared to GPT-4o's 13%. Despite its advanced capabilities, the o1-preview model lacks some practical features present in GPT-4, such as web browsing and file uploading. However, Sam Altman-backed OpenAI emphasizes that the model's strength lies in tackling complex, multi-step tasks, making it especially useful for industries requiring high-level problem-solving. Altman highlighted this shift in AI performance, saying, "It is the beginning of a new paradigm: AI that can do general-purpose complex reasoning." OpenAI also introduced a smaller, more affordable version called o1-mini. It is tailored specifically for developers needing advanced coding capabilities without extensive world knowledge. Moreover, this model is 80% cheaper than the o1-preview version, making it a cost-effective solution for developers. From today, users of ChatGPT Plus and Team can manually select o1-preview and o1-mini from the model picker. The o1-preview model comes with a rate limit of 30 messages, while o1-mini offers a 50-message limit. API users in the highest usage tier can also begin using the models. However, certain features, such as function calling and streaming, are not yet available. Meanwhile, as part of its focus on safety, Sam Altman's OpenAI implemented new security training approaches for these models. In jailbreak tests, the o1-preview outperformed GPT-4o, scoring 84 out of 100 compared to GPT-4o's 22. Furthermore, the company has also expanded its partnerships with AI safety institutes in the U.S. and U.K. to further strengthen its safety measures. Looking ahead, OpenAI plans to expand access to o1-mini for ChatGPT Free users. In addition, the company will continue adding new features to the o1 series, including browsing and file uploads. Sam Altman concluded his announcement by acknowledging the models' imperfections but expressed optimism about their potential. He stated, "o1 is still flawed, still limited, and it still seems more impressive on first use than it does after you spend more time with it. but also, it is the beginning of a new paradigm." Whilst, OpenAI is now eyeing a $150 billion valuation with its latest funding.

OpenAI unveils o1-series AI models: What are they, how they work, and more: Technology news

These capabilities will be useful for tackling complex problems in science, coding, maths, and similar fields. To test this, OpenAI prompted the reasoning model to solve the qualifying exam for the International Mathematics Olympiad (IMO). Compared to the GPT-4o model, which correctly solved 13 per cent of the problems, the new o1 model answered 83 per cent correctly. OpenAI o1 and o1-mini: Differences The OpenAI o1-mini AI model is described as a faster, more cost-effective reasoning model with greater efficiency in coding tasks. Being smaller than the o1 model, the o1-mini is 80 per cent cheaper, making it a more efficient solution for applications requiring reasoning but not "broad world knowledge."