OpenAI is gearing up to introduce “Strawberry,” an AI model with a focus on reasoning, into ChatGPT within the next two weeks. Strawberry, currently a code name, represents the upcoming AI model possessing the capability to tackle complex problems through in-depth internet research, a feature yet to be seen in any existing AI model. This move reflects OpenAI’s ambition to enhance ChatGPT’s abilities by incorporating this powerful new service.
Strawberry: The Reasoning AI
Strawberry, a novel AI model on the horizon, stands out for its capacity to think before providing information, earning it the label of “reasoning-focused artificial intelligence” due to its knack for solving intricate problems. The wait for its release won’t be long; according to a recent report in The Information, OpenAI has plans to launch Strawberry within the coming two weeks.
Strawberry’s Unique Features
Thoughtful Responses: Strawberry distinguishes itself from other conversational AI models by its ability to pause and reflect before responding, rather than offering an immediate answer, as noted in The Information’s report. This approach enables Strawberry to generate more contemplative and well-considered responses.
Text-Only Interaction: The report also indicates that the new AI model will interact exclusively through text prompts and responses. Unlike the current ChatGPT version, it won’t have the capability to interpret images. While this limits its multimodality, it also reduces the likelihood of hallucinations.
Complex Problem-Solving: Strawberry is poised to be a groundbreaking reasoning model, equipped to handle complex, multi-step requests that demand advanced problem-solving skills. This encompasses a wide range of tasks, from tackling challenging algebraic equations to crafting intricate, long-term marketing strategies.
Response Time and Cost: The Information highlights that Strawberry’s response time can vary between 10 to 20 seconds, considerably longer than other conversational AI models. Despite being designed for quick responses to simple questions, it can sometimes linger on straightforward queries. Early testers found the marginally improved responses didn’t justify the extended wait times.
Compute-Heavy and Rate Limits: These delays likely stem from Strawberry’s advanced capabilities. It’s expected to be more computationally demanding than its counterparts and, consequently, more expensive to use. Reports suggest potential rate limits restricting users to a maximum number of messages per hour, with the possibility of a higher-priced tier offering faster responses.
Beyond Strawberry: GPT-Next
OpenAI’s development efforts extend beyond Strawberry, encompassing projects like Orion and more. Earlier this month, OpenAI Japan CEO Tadao Nagasaki revealed during the KDDI Summit 2024 that the next AI model, GPT-Next, is on the horizon, promising to be 100 times more powerful than GPT-4.
Enhanced Performance: GPT-Next will reportedly leverage a smaller version of OpenAI’s secretive Project Strawberry (previously known as Project Q* or Q-star). Its improved performance stems not solely from increased computational power but from refinements in its underlying architecture, enabling more efficient and effective processing.
Add Comment