“Especially when we just saw GPT-5 (ultimately relabeled as GPT 4.5) fail despite more than two years of hard and expensive work.”
Our reading, the sentence that was checked:
OpenAI's planned GPT-5, relabeled GPT-4.5, failed despite more than two years of hard and expensive work.
Assessment: Unresolved Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Ezra Klein’s new take on AGI – and why I think it’s probably wrong, paragraph 27.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-ezra-kleins-new-take-on-agi-and-why-1dcb61-a1
We recorded 29 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-28. The source may have been edited after 2025-03-05. This page does not claim the wording was captured then.
What the claim commits to
- Kind
- Observation
- Deadline, as the speaker wrote it
- None stated. This site does not add one.
- Deadline state when assessed
- No horizon
- Deadline state on 2026-10-29
- No horizon (the release’s reference date)
Conditions
- despite more than two years of work (qualifier, stated by the speaker)
From the speaker
- GPT-5
- ultimately relabeled as GPT 4.5
- fail
- more than two years
- hard and expensive work
Our interpretation
- Read this as a conjunction: the project intended as GPT-5 became GPT-4.5, missed intended capability goals, and involved more than two years of costly development.
- Failure ordinarily means missing the intended capability advance in this context. It does not mean no product shipped or no improvements occurred.
- More than two years is a duration claim, not a forecast condition.
- The described period is the project’s development through its recent release. The post supplies no exact start date.
- Expensive is a qualitative spending assertion without a currency, accounting basis, or amount. Those details are not inferred from unrelated company spending.
Counts as supported if: Reliable project records or corroborated reporting establish the original intended identity, subsequent GPT-4.5 release, missed capability goals, development exceeding two years, and substantial project expenditure.
Counts as contradicted if: Reliable project records refute an essential component, such as the intended identity, duration exceeding two years, or failure to meet the original capability goals.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Unresolved
The record was checked, and the evidence available by the cutoff does not settle it.
Scored on a plain reading. The speaker named no measure for this claim, so it was judged on the reading an informed reader would take, stated below under our interpretation. Both reviews had to find that reading fair.
- Evidence as of
- Deadline state when assessed
- No horizon
- Assessment recorded
- Policy version
- 2026-10-28.7
Rationale
Project reporting and Altman’s reproduced announcement support the connection between Orion, the intended GPT-5 project, and GPT-4.5. Reporting also describes missed expectations, while OpenAI documents actual improvements. The record does not establish the exact claim of more than two years of development, original capability targets, or a defined spending measure. The full conjunction therefore remains unresolved.
Limitations
- Product improvements can coexist with missed internal expectations.
- Reporting of more than 18 months is a lower bound, not proof that development lasted less than two years.
- The qualitative spending assertion has an unknown accounting basis and cannot receive a decisive financial score.
- The Altman statement is available here through a secondary reproduction.
Evidence (3 items)
Context only: The Wall Street Journal, republished by Mint,
came out in March
Why it matters here: The report identifies Orion as the GPT-5 project, describes development exceeding 18 months, and reports training results below researchers’ expectations using people close to the project. Its chronology does not establish more than two years by March 5, 2025, or exclude earlier preliminary work.
- Publisher
- The Wall Street Journal, republished by Mint
- Kind of source
- News report
- Describes events of
- Public from
Context only: Simon Willison’s Weblog,
called Orion internally
Why it matters here: This dated reproduction of Altman’s statement directly links Orion with GPT-4.5. It does not establish Orion’s original intended public name, development duration, expenditure, or whether it missed its original targets.
- Publisher
- Simon Willison’s Weblog
- Kind of source
- Secondary summary
- Describes events of
- Public from
Context only: OpenAI,
a step forward
Why it matters here: OpenAI released GPT-4.5 and reported improvements over GPT-4o across the appendix’s listed evaluations. This counters a reading of failure as no improvement, but does not show that the original internal targets were met.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
What would settle it
Dated project records establishing the start, intended name and capability targets, release decision, and project-specific expenditure, or corroborated reporting that establishes those details.
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Ezra Klein’s new take on AGI – and why I think it’s probably wrong
Especially when we just saw GPT-5 (ultimately relabeled as GPT 4.5) fail despite more than two years of hard and expensive work.
- Where
- paragraph 27
- Published
- 2025-03-05
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-ezra-kleins-new-take-on-agi-and-why-1dcb61-a1: Current. Outcome: Unresolved. Evidence as of 2026-10-28. Public date not recorded.
Recorded events
Assessment publishedBy this site
First assessment published after both reviews and the final-text check passed.
About assessment marcus-ezra-kleins-new-take-on-agi-and-why-1dcb61-a1.