OpenAI Shares Benchmark Scores of o3 Series AI Model
OpenAI has recently announced the benchmark scores of its new o3 series of artificial intelligence (AI) models. This series is the successor to the previously released o1 models. OpenAI claims that the o3 series significantly outperforms its predecessor in various tasks. The o3 series includes two models: the standard o3 and the smaller o3-mini. These models are designed for advanced reasoning tasks, showcasing the company’s commitment to enhancing AI capabilities. Currently, OpenAI is conducting public safety testing for these models, with plans to launch the o3-mini to the public early next year. This announcement comes shortly after Google introduced its Gemini 2.0 Flash Thinking Mode AI model, highlighting the competitive landscape in AI development.
OpenAI Begins Public Safety Testing of o3 Series AI Models
On the twelfth day of its shipping schedule, OpenAI made a significant announcement regarding the o3 series. During a live stream, CEO Sam Altman explained that the company decided to skip the name o2 due to a trademark conflict with the UK telecommunications provider Telefonica. Instead, the new models are branded as o3 and o3-mini. Although the announcement generated excitement, OpenAI clarified that the models are not yet available for public use. Instead, they are being provided in early access to selected external researchers for public safety testing.
Researchers interested in participating in this testing phase can apply through OpenAI’s platform until January 10. The company emphasized that both the o3 and o3-mini models will offer enhanced performance compared to the o1 series. They are expected to tackle more complex tasks in areas such as coding, mathematics, and natural language processing. However, OpenAI has not disclosed the full range of capabilities for these new models. During internal testing, the o3 model achieved impressive scores on benchmark evaluations, scoring 71.7 percent in the SWE-bench benchmark and 96.7 percent in the AIME 2024 benchmark. These results indicate a significant improvement over the o1 series, but a comprehensive evaluation will only be possible once the models are publicly available. The o3-mini is anticipated to launch in January 2025, further expanding OpenAI’s offerings.
OpenAI Provides Unlimited Access to Sora for Paid Subscribers
In a separate announcement, OpenAI’s CEO Sam Altman revealed a special offer for ChatGPT Plus and Teams subscribers. As a “day-13 bonus,” these subscribers will receive unlimited access to Sora during the last two weeks of December. Altman explained that this initiative is made possible due to reduced server loads as many offices close for the Christmas holiday. This offer aims to enhance user experience and provide additional value to subscribers during the festive season.
Rohan Sahai, a product lead at OpenAI, added details about the upgrades to Sora. He mentioned that the company has improved Sora’s blend feature, allowing for more seamless integration of AI-generated content. Additionally, users can now share links to their AI-generated videos with friends, even if those friends do not have an account. This feature encourages sharing and collaboration, making Sora more accessible and user-friendly. Overall, OpenAI’s initiatives reflect its commitment to enhancing user experience while advancing the capabilities of its AI models.
Observer Voice is the one stop site for National, International news, Editor’s Choice, Art/culture contents, Quotes and much more. We also cover historical contents. Historical contents includes World History, Indian History, and what happened today. The website also covers Entertainment across the India and World.
Follow Us on Twitter, Instagram, Facebook, & LinkedIn