OpenAI said 80 to 90 percent of its research already targets GPT 7, GPT 8, and beyond, according to Boris Power, the company's Head of Applied Research. Power stated that most of the value in AI development comes from major model generations rather than incremental updates within a single series.
Improvements within a generation, like going from GPT 5.1 to 5.2, come from specialized training data but are intentionally short-term bets, Power said. The real leap, however, comes with a new model generation, where 'everything else just works a lot better,' he added.
Power noted that incremental updates are 'seen as extremely shortsighted' within the company but help OpenAI iterate and learn faster today, although they are not the right long-term strategy. These updates allow the company to test and refine ideas quickly before committing to larger, more complex projects.
"The main challenge with today's AI assistants is not model quality but onboarding," Power said. 'Most ChatGPT users don't know what they can do with AI,' he added, suggesting future models should better surface what's possible and anticipate user needs.
The announcement follows a period of increased focus on long-term AI development, as OpenAI seeks to address user limitations in current AI tools. Power emphasized that while incremental improvements are useful, they are not the primary focus of the company's research efforts.
OpenAI did not say when GPT 7 will be released, and it raised the question of how to balance short-term gains with long-term innovation. Power mentioned that the company is still figuring out where to invest for quick wins after each major model update.
Source: thedecoder
