REFUTED
The most important thing and it sounds like the most boring thing I can say... the models are just going to get smarter, generally across the board,
Source: The Verge ·
Analyst note
This entry tracks a structural feature of AI markets: executive language about monotonic capability improvement is easy to agree with, while the tradable claim—the calendar and magnitude of a numbered flagship release—is not. Altman’s Build quote is cautious and (literally) non-committing, yet the surrounding ecosystem repeatedly converted it into an imminent GPT-5 date.
The refuted element is the market hypothesis, not necessarily the quoted sentence. OpenAI’s incentives favor shipping when marginal reliability clears internal gates; commentators’ incentives favor headline milestones. Models tied to the claim include GPT-4-era baselines and the eventual GPT-5 reception. What to watch after release is whether evaluations show discontinuous capability jumps versus tighter packaging of prior competencies, and whether enterprise buyers change contracts based on the new SKU.
Evidence timeline
At Microsoft Build 2024, Altman teased ongoing progress toward the next major OpenAI model generation; press coverage widely interpreted this as imminent GPT-5-class shipment within 2024.
Leak- and analyst-driven timelines continued to cluster around late-2024 or early-2025 for a flagship successor, with OpenAI publicly emphasizing quality bars over release clocks.
Through 2025, OpenAI shipped strong mid-cycle upgrades but did not release a GPT-5-branded flagship in the window many traders priced; ‘smartness’ improved but naming and packaging splintered.
A GPT-5-class flagship did not ship in calendar 2024; mid-2026 reviews framed the eventual release as meaningful but not aligned to the most aggressive end-of-2024 expectations.