OpenAI says most research now targets GPT-7 and beyond
Boris Power, OpenAI’s head of applied research, says 80 to 90 percent of the company’s research is directed toward GPT-7, GPT-8 and later generations. He said this is where OpenAI expects most of the value to emerge. By contrast, changes within one generation, such as moving from GPT-5.1 to GPT-5.2, rely on specialized training data and are treated as short-term investments.
Power described major gains as coming with a new model generation, when broader improvements become possible. After each such step, OpenAI must reassess which areas can produce rapid benefits. Although incremental updates are considered extremely short-sighted internally, he said they help the company iterate and learn more quickly in the near term, even if they are not the best long-term approach.
According to Power, the main limitation of current AI assistants is user onboarding rather than model quality. Many ChatGPT users do not know what the systems can do. He expects future models to explain their capabilities more clearly and anticipate user needs. In his comparison, GPT-4 required careful prompting, GPT-5 needs substantial feedback, while GPT-6 can be given a goal more like a capable colleague.
Written by Sakhanda Wire AI from reporting by The Decoder. Facts, quotes and full context are in the original article below.