OpenAI DevDay won’t feature a new GPT model

While last year’s OpenAI event in San Francisco was a headline-grabbing spectacle, introducing several new initiatives including the now-defunct GPT Store, this year’s setup is markedly toned down. Instead of one grand event, OpenAI plans to connect with developers through multiple, localized sessions as part of their reimagined DevDay conference. This year, the spotlight shifts away from launching major new models to refining and expanding the existing suite of API and developer tools.

This year, OpenAI’s DevDay will go global with events scheduled in San Francisco on October 1st, London on October 30th, and Singapore on November 21st. These gatherings will showcase a mix of workshops, breakout sessions, and live demonstrations featuring OpenAI’s product and engineering teams. Highlighting the talent of developers, the event also offers spotlights on their work. Entry is priced at $450, though scholarships are available for qualifying participants, with applications closing on August 15th.

At our global DevDays, you can look forward to:

• Workshops: Hands-on technical sessions to deepen your skills and explore new possibilities with the OpenAI platform and API tools.
• Breakout sessions: Discussions on best practices around model customization, evals, steerability, scaling, and a variety of other topics, led by AI experts and community members.
• Demos: Meet the OpenAI product and engineering teams and see our models and platform in action.
• Developer spotlights: Take a look at boundary-pushing projects created by our talented developer community and top startups.
• Evening reception: Close out the evening with dinner, drinks, and entertainment.

-OpenAI

Recently, OpenAI has adopted a more gradual approach to innovation in generative AI, focusing on refining its current tools rather than making groundbreaking advancements. The company is diligently working on the successors to its GPT-4o and GPT-4o mini models, emphasizing enhancements in model stability and performance. However, despite these efforts, OpenAI is perceived by some as having relinquished its lead in the generative AI field, potentially due to the hurdles in sourcing high-quality training data.

The challenge is compounded as more content creators restrict access to their online materials, concerned about plagiarism and lack of proper attribution or compensation.

If the trend of restricting data access persists, experts forecast a potential data drought for training generative AI models in the upcoming years. This predicament, combined with the threat of copyright litigation, has compelled OpenAI to forge expensive licensing deals with publishers and data brokers.

In a promising development, OpenAI has crafted a new reasoning method aimed at enhancing model performance, especially in solving mathematical problems. Mira Murati, the company’s CTO, has hinted at a forthcoming model endowed with “Ph.D.-level” intelligence. This revelation was part of a May blog post announcing the initiation of their next advanced model project, setting high expectations and mounting pressure on the company. OpenAI’s financial strategy involves substantial expenditures on model training and recruiting elite research talent, costing the company billions.

Despite these advancements, OpenAI continues to navigate a host of controversies—from the use of copyrighted data and stringent non-disclosure agreements for employees to the marginalization of safety researchers. The shift towards a slower product development cycle could, however, help alleviate concerns, suggesting a recalibration of priorities towards enhancing AI safety amidst the drive to develop more sophisticated AI technologies.

Featured image credit: Jonathan Kemper/Unsplash