OpenAI's Boris Power: 80 to 90 percent of research targets GPT 7 and beyond

Boris Power, OpenAI's Head of Applied Research, has laid out a clear split in how his company thinks about progress. The small step from GPT 5.1 to 5.2 comes from specialized training data, but Power describes these moves as intentional short-term bets, and he says they are viewed internally as "extremely shortsighted." Their value is speed of iteration, not the destination.

The destination, in Power's framing, is a new generation such as GPT 6, GPT 7, or GPT 8, where "everything else just works a lot better." He notes that after each generational jump, the company has to relearn where to invest for quick wins — a reminder that even OpenAI does not get a permanent playbook from its own breakthroughs. Against that backdrop, Power argues the bottleneck for today's assistants is not model quality but onboarding — most ChatGPT users do not know what the tools can do. He traces the arc from GPT-4, which required careful prompting, to GPT-5, easier to use but still feedback-hungry, to GPT-6, which he describes as a capable colleague you can simply hand a goal to.

The stakes hinge on whether that onboarding gap is really the binding constraint, since Power's own account suggests generational jumps — not user education alone — are where the leap comes from. For teams building on OpenAI's models, the practical question is whether to squeeze more out of current versions or wait for the next generation; Power's comments suggest the company is betting heavily on the latter.