OpenAI has directed 80–90% of its research toward GPT-7, GPT-8, and...

OpenAI has directed 80–90% of its research toward GPT-7, GPT-8, and future generations

Boris Power, Head of Applied Research at OpenAI, stated on September 23, 2026, at the Fellows Forum 2026 session in Menlo Park, that approximately 80–90% of the company's research work is already focused on GPT-7, GPT-8, and subsequent generations, while OpenAI considers improvements within a single generation to be short-term bets.

Compiled by Sergey KostenchukPublished 2026-09-28Updated 2026-09-28

2026-09-28 Research OpenAI

🤖 OpenAI: 80–90% of research is already focused on GPT-7 and GPT-8

Boris Power, Head of Applied Research at OpenAI, told the Fellows Forum 2026 (September 23, Menlo Park) that 80–90% of the company's research is already going toward GPT-7, GPT-8, and future generations. He called within-generation updates like GPT-5.1 → 5.2 short-term bets.

🌍 The leap in value comes from generational shifts, not point releases. In the GPT-6 lineup, the flagship Astra ($10/$50 per 1M tokens) serves as a "teacher" for the cheap Luna ($0.10/$0.50, a 100x difference); Sol is the mid-tier.

👤 Don't expect breakthroughs from intermediate versions: quality leaps happen at the GPT-7/GPT-8 boundary, with no dates. According to Power, most ChatGPT users don't know what AI is capable of — future models should suggest their own capabilities.

Source 1: https://the-decoder.com/openai-says-80-to-90-percent-of-its-research-already-targets-gpt-7-and-beyond/

Source 2: https://www.youtube.com/watch?v=Xzsg-rLGzas

Sources