OpenAI is about to release a new AI model codenamed "Strawberry" in the fall, aiming to significantly improve reasoning capabilities and become a powerful data source for GPT-5. The model promises to solve more complex problems, including never-before-seen mathematical challenges and subjective issues that require deep thinking, such as marketing strategies. Its powerful language processing capabilities can even solve complex word puzzles, showing broad application potential beyond the technical field.
At a time when global AI technology competition is becoming increasingly fierce, OpenAI’s move will undoubtedly once again lead the industry trend. The "Strawberry" model will not only be integrated into ChatGPT, but will also be used to generate synthetic data for training the "Orion" model, solving the problem of high-quality data acquisition, and is expected to bring more revenue to OpenAI and promote rapid business growth. Although OpenAI is currently facing losses, its high valuation and market expectations for "Strawberry" indicate that its influence in the AI field will further increase. However, "Strawberry" and the future GPT-5 have also triggered discussions about the potential risks of AI, and OpenAI emphasized that its goal is to use AI to better serve mankind.
OpenAI plans to integrate "Strawberry" into ChatGPT and is seeking additional financial support to achieve this goal. This move is expected to bring more revenue opportunities to OpenAI and drive rapid growth of its business.
Although OpenAI may currently be losing more money per month than revenue, its private valuation has reached $86 billion. The market's expectations for "Strawberry" have also risen.
Strawberry will also be used to generate synthetic data for training the Orion model. This method of data generation helps OpenAI overcome the difficulty of obtaining high-quality data.
Alex Graveley, CEO of Minion AI, believes that using high-quality training data generated by "Strawberry" will help reduce errors produced by the model, the so-called "hallucinations."
This summer, OpenAI demonstrated Strawberry’s capabilities to national security officials, demonstrating its potential for national security. Some people worry that if GPT-5 Orion is released, it may control the entire Internet and become a new tool for surveillance and propaganda. But OpenAI's goal is clear - to use AI to provide better services to humans, not to replace humans.
Reference: https://www.theinformation.com/articles/openai-shows-strawberry-ai-to-the-feds-and-uses-it-to-develop-orion
The release of the "Strawberry" model heralds that artificial intelligence technology will enter a new stage of development, and its application prospects are worth looking forward to. It also puts forward higher requirements for the ethics and safety of AI technology. The future development of OpenAI will largely depend on how to balance technological progress and social responsibility.