On Tuesday, September 10, according to The Information, a well-known technology and business media, citing multiple people who have tested the model, OpenAI, the "AI darling" standing at the forefront of the times, plans to release the latest artificial intelligence called "Strawberry" in the next two weeks. Smart products as part of the services provided by ChatGPT. That's earlier than the "fall release" (September to November) that was rumored two weeks ago.
The above report stated that there are still some urgent problems that need to be solved before and after the release of the artificial intelligence "Strawberry" that focuses on reasoning capabilities, but its seemingly "hurried launch" seems to indicate that OpenAI has felt the fierce competitive pressure in the field of large language model-driven products, and hopes to use the recent The strawberry project, which has been long-awaited by the market for several months, will restore the momentum that ChatGPT has enjoyed for most of the two years since its launch.
Two weeks ago, there was news that OpenAI accelerated the launch of the "Strawberry" reasoning artificial intelligence model to promote the chat robot business. It was reported four weeks ago that the CEO of perplexity, a well-known AI search company, hinted that it had used OpenAI’s latest model “Strawberry” in advance.
Wall Street News mentioned that as early as May, there were rumors that OpenAI was mysteriously developing a project code-named "Strawberry". Two months ago, on July 13, there was news that "Strawberry" was a new artificial intelligence model being developed by OpenAI, hoping to provide more advanced reasoning capabilities to form a major breakthrough towards general artificial intelligence (agi).Specifically, with its enhanced reasoning capabilities, the "Strawberry" model can better handle complex scientific and mathematical problems that have made previous AI models "scratching". It not only allows large language models (LLM) to generate answers based on users' complex queries, You can even plan in advance to browse the Internet autonomously and reliably and conduct "deep research" as defined by OpenAI.
Commentators generally believe that OpenAI ’s “Strawberry Project” is expected to become the next major breakthrough in artificial intelligence. More advanced reasoning capabilities mean the technology can perform more complex tasks "autonomously" rather than just relying on step-by-step prompts from the user. Reasoning has always been one of AI's toughest problems in the pursuit of human-like intelligence, and AI models with reasoning capabilities can not only improve mathematics and programming capabilities, but also revolutionize scientific research and enhance decision-making in complex fields such as healthcare and finance. strength and accelerate technological innovation.
It is reported that although "Strawberry" is released as a part of ChatGPT, it is also an independent product and may be included in the drop-down menu of AI models selected by customers to support the operation of ChatGPT . However, the specific provision form and pricing model are still uncertain.
Unlike regular conversational AI, "Strawberry" will "think" for 10 to 20 seconds before responding to the user's prompts, rather than giving an answer immediately. For queries that are complex or require multiple execution steps, "Strawberry" will be easier to use than the existing GPT-4o version, and does not require users to explicitly tell ChatGPT to "complete the intermediate reasoning steps to arrive at the final answer" and other additional actions.
Some analysts say that this means that "Strawberry" will become a more intelligent assistant, not only better at solving mathematical problems and coding, but also better at more "subjective" business tasks, such as brainstorming product marketing strategies. And in such creative tasks, the model will provide more detailed suggestions that are more specific to the user's company needs, such as generating a weekly execution plan.
However, assuming everything goes well with the release, in the initial version "Strawberry" can only receive and generate text information, but not images. This means that it does not have the "multi-modal" function like other OpenAI models, which is one of the "shortcomings" .
At the same time, although the "thinking" stage helps avoid AI mistakes and can predict how to comprehensively answer user questions, some people who have tested the model said that some simple questions will also make Strawberry "think" and fail to give answers quickly; The quality of the answers is only slightly better and not worth the extra wait; and there is sometimes a failure to remember and integrate previous chats with the user before answering a new question.
In terms of pricing, there was news last week that the current paid price of ChatGPT Plus is US$20 per month, but OpenAI is considering drafting new large models such as "Strawberry" and "Orion", which is known as "ChatGPT 5". A monthly payment of US$2,000 is equivalent to a 100-fold increase in pricing. Analysts believe that OpenAI can consider such a high price span mainly because of lack of money, strong enough new products, and hundreds of millions of users.OpenAI CEO Sam Altman said earlier this year that "the most important direction of progress in the field of artificial intelligence will be reasoning capabilities."
The "Strawberry" project is considered to be the OpenAI "Q*" plan disclosed last year. The project has been regarded as a breakthrough within the company. People who watched the demonstration said it was able to answer tough scientific and mathematical questions that current commercial AI models cannot solve.
OpenAI gave a five-stage roadmap for future AI development on August 11. The first stage is the current AI chatbot with conversational language, and the second stage is the "reasoner" with human-level problem-solving capabilities. The third stage is the "agent" (ai agent) who can take autonomous actions, followed by the "innovator" who can assist in invention and the "organizer" who can complete the organizational work.
" Strawberry" is likely to be the key to helping OpenAI achieve second-level AI . Once successful, it will redefine the boundaries of AI's capabilities , autonomously perform complex tasks, independently solve complex problems, and verify output content, representing human beings one step closer to general artificial intelligence (AGI). . "Strawberry" focusing on reasoning may also be the only way to lay the foundation for the next generation AI model "Orion" or ChatGPT 5.
Ghazenfer Mansoor, founder and CEO of Technology Rivers, a software development agency, pointed out that with the "Strawberry" project, OpenAI's pursuit of human reasoning levels is not only a technical miracle, but also able to break boundaries and inspire new possibilities in every field. For example, in the business field, artificial intelligence will be able to greatly change the way supply chains are managed, predict market trends very accurately, and make customer experiences very personalized to a large extent.
At the level of ordinary users, the user experience will also be improved. For example, AI models with reasoning capabilities can not only provide more optimized travel plans and health recommendations, but can even analyze users’ consumption habits and create fully customized budget plans. , become a smarter assistant.
All in all, in an ideal scenario, Strawberry will be able to solve problems and tasks beyond the capabilities of current AI models and conduct "deep dives" that "outperform any existing chatbot."
Other analysts pointed out that the "Strawberry" plan aims to raise more funds, which OpenAI needs to support the development of the next generation cutting-edge model code-named "Orion". Moreover, generating high-quality synthetic training data for the Orion model is considered one of Strawberry's key applications , "which is significant because most of the training data on the Internet has been exhausted":
The large data sets on which AI models rely can be prone to bias and error, or contain incomplete or inaccurate information. The high-quality synthetic data generated by Project Strawberry can fill gaps in real-world datasets and provide more comprehensive, inclusive, and balanced training sets.
Many people believe that using synthetic data can help future artificial intelligence models be more neutral and fair, and reduce noise, hallucinations, and irrelevant information, thereby improving the efficiency of training and the accuracy of the model.
Project Strawberry improves reasoning, logic, and the ability to plan and conduct research based on what is known, allowing models to autonomously conduct experiments, analyze data, and formulate new hypotheses. This could lead to scientific breakthroughs, including the discovery of new drugs. These models can also provide personalized education and create educational content and interactive courses.
According to people familiar with the matter, the "Strawberry" project includes a specialized "post-training" method, that is, after the generative AI model has been pre-trained on a large data set, the model is further adjusted to improve its performance on specific tasks. This is similar to the "Self-Taught Reasoner" (STaR) method developed by Stanford University in 2022.
Noah Goodman, one of the creators of STaR and a professor at Stanford University, once said that STaR allows AI models to iteratively create their own training data and "guide" themselves to a higher level of intelligence. In theory, it can be used to allow language models to achieve intelligence beyond humans. . But he also admitted, "This is both exciting and scary, and humans need to think seriously about their situation."