OpenAI plans to launch a new model codenamed 'Strawberry' this fall, aimed at significantly enhancing reasoning capabilities and serving as a source for generating data to support GPT-5. This model excels at solving complex problems, including mathematical challenges and subjective issues like product marketing strategies, and demonstrates strong capabilities in language tasks. 'Strawberry' will be integrated into ChatGPT, as OpenAI seeks additional funding to achieve this, which is expected to create revenue opportunities for the company and drive business growth. Although OpenAI currently...

OpenAI co-founder Sam Altman shared a photo of strawberry growth on social media platform X, captioning it 'I love summer in the garden,' sparking speculation in the AI community about a new foundational model codenamed 'Strawberry' within OpenAI. This model could be comparable to GPT-5, aiming to showcase advanced reasoning capabilities and even the ability to browse the internet and conduct deep research. This move suggests that OpenAI may be exploring general AI intelligence, alongside companies like Anthropic, Google, and Meta.

OpenAI's CEO, Sam Altman, recently shared his insights on the highly anticipated GPT-5 model during an interview at the Aspen Ideas Festival, drawing widespread industry attention. Despite many users expecting OpenAI to unveil GPT-5 during its Spring Update event, the company eventually released the new flagship model, GPT-4o. Regarding the development progress of GPT-5, Altman said, "We still don't know. We are optimistic, but there is still much work to be done." Although the specific release

Recently, during an interview at the Aspen Ideas Festival, OpenAI CEO Altman candidly shared the progress and challenges in the development of GPT-5. He emphasized that despite the team's confidence in the project, the path to GPT-5 is still long and fraught with unknowns. "We are currently optimistic, but the development work remains arduous and challenging," he said, revealing a desire for technological breakthroughs tempered with caution.Regarding the specific release date of GPT-5, Altman di

OpenAI CEO reveals that the release of GPT-5 will bring significant advancements. GPT-5 is expected to be released in mid-2024 and is currently in the training phase. GPT-5 will be a large language model, offering smarter text prediction capabilities. Enterprise clients have received demo versions of GPT-5, with an official launch anticipated this summer.

OpenAI is offering clients a preview version of GPT-5, with red team testing ongoing, and a release expected this summer. Business users are experiencing the latest version of ChatGPT, which shows a significant leap in performance, enabling the model to utilize AI agents to complete tasks. Altman revealed that GPT-5 will see a comprehensive upgrade, necessitating regulatory bodies to manage AI to prevent individuals or companies from having full control over AGI. The release of GPT-5 is expected to drive revenue growth for OpenAI, with a substantial enhancement in model performance, and Altman stated it will change human history.

OpenAI plans to release GPT-5 this summer, with significant performance enhancements expected to drive revenue growth. Altman hints that OpenAI's products will change human history, with GPT-5's performance exceeding expectations. GPT-5 is poised to impact various fields and business sectors, and Altman firmly believes that building AGI beyond human capabilities is not difficult. Altman warns that startups underestimating GPT-5 will be 'crushed' by the next generation of models. He states that many startups that underestimate GPT-5 could find themselves disadvantaged in the future technological upheaval.

ChatGPT has quietly updated a major feature, allowing users to @ any GPTs in the GPT Store during conversations, similar to mentioning someone in a group chat. Users have found that this update changes their interaction with GPTs, allowing for a shared context between different GPTs and opening new workflows. The number of GPTs in the GPT Store has exceeded three million, enabling users to combine different GPTs for more efficient task handling. GPT developers are also inspired, believing that...

The OpenAI Board announced it has veto power, particularly focused on the safety risks of cutting-edge models such as GPT-5. The company has established a security advisory team that reports monthly to ensure management is aware of misuse cases for models like ChatGPT. Under the new safety framework, the company has set requirements for models to achieve safety scores before entering the next development phase. Three safety teams have been formed to address AI risks across different time scales, mainly covering cybersecurity and CBRN risks. The company conducts regular safety drills and invites third-party red teams for independent evaluations of the models to ensure their safety.

["B100's speed could exceed H100's by more than 4 times, featuring a chip combination design", "GPT-5 may achieve general artificial intelligence, requiring substantial computational resources", "Microsoft has invested over $10 billion into OpenAI", "NVIDIA's B100 and OpenAI's GPT-5 are anticipated to launch in 2024", "Industry experts predict both will lead to significant improvements in computing speed and intelligence"]

OpenAI is partnering with Microsoft to initiate the development of GPT-5. The goal of GPT-5 is to create a super AI that matches human intelligence. GPT-5 is likely to require more data support and is seeking large-scale datasets from various sources. The technical challenges facing GPT-5 include dependencies on chips, among others. OpenAI's CEO stated that the specific functionalities of GPT-5 are currently difficult to predict.

OpenAI founder Sam Altman revealed the company's multiple plans and outlook for the future in an FT interview. The collaboration with Microsoft is going smoothly, and the company plans to raise more funds to address the costs of building complex AI models. Updates to the business model include GPT-4 upgrades and new tool releases, focusing on custom versions of ChatGPT and a GPT store. OpenAI is actively developing the next generation AI model, GPT-5, calling for large-scale datasets. The company emphasizes the pursuit of Artificial General Intelligence (AGI), building stronger autonomous agents and creativity.

The University of California, Santa Cruz has developed the MiniGPT-5 model, which achieves effective alignment between text and images. MiniGPT-5 utilizes a multimodal encoder and Generative Vokens technology to generate coherent, high-quality text and images. The model outperforms comparative baselines in tests across multiple datasets, demonstrating strong adaptability.

["🤔 Generative AI Has Reached Its Peak: Bill Gates expresses concerns about the development of generative artificial intelligence.", "📉 Uncertain Outlook for GPT-5: Gates believes that GPT-5 will not have significant improvements in the future.", "💡 Short-term Potential of AI: Gates is optimistic about the development of artificial intelligence in the short term.", "💼 Role of AI in Healthcare: Gates discusses the potential of artificial intelligence in the medical field."]

Despite OpenAI's expectation that GPT-5 will show significant improvements over GPT-4, Bill Gates remains skeptical. Gates believes that the advancements from version 2 to 4 have been remarkable, and he doubts that there will be much enhancement in version 5. However, he is optimistic about the current AI systems' potential applications in fields like drug development. Gates believes that explainable AI can be developed over the next decade. As an investor in OpenAI, Microsoft is also closely monitoring its technological advancements.

OpenAI has filed two trademark applications for 'GPT-5' in China, categorized under Class 9 and Class 42. GPT-5 will offer functionalities including text generation, natural language understanding, speech transcription, translation, and analysis. OpenAI has stated that training for GPT-5 has not yet begun, and there is still much work to do.

OpenAI has unveiled their 'official crawler' GPTBot, which is used to train AI models like GPT-4 and GPT-5. This allows website owners to restrict its crawling of content as needed, thereby improving the accuracy and safety of AI models.