“In the last year, his company has released no major update to GPT-4, and many competitors have caught up or are close. Competitors are eroding the initial gap in quality and starting to offer similar products, sometimes at significantly lower prices.”
Our reading, the sentence that was checked:
In the last year OpenAI has released no major update to GPT-4, many competitors have caught up or are close, and competitors are eroding the initial gap in quality and starting to offer similar products, sometimes at significantly lower prices.
Assessment: Mixed Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, The Sam Altman Playbook, paragraph 52.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-the-sam-altman-playbook-6221c7-a1
We recorded 18 claims from this post. See every claim from this post.
The wording above is the version this site observed on 2026-10-29. The source may have been edited after 2024-05-05. 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
- in the last year (scope, stated by the speaker)
- many competitors (scope, stated by the speaker)
- sometimes at significantly lower prices (qualifier, stated by the speaker)
From the speaker
- In the last year
- no major update to GPT-4
- many competitors have caught up or are close
- eroding the initial gap in quality
- similar products
- sometimes at significantly lower prices
Our interpretation
- The preceding year runs from May 5, 2023, through May 5, 2024.
- A major update means a substantial improvement to GPT-4 capabilities or usable functionality, including context capacity and supported inputs. It need not be a newly numbered model generation.
- Caught up or close means multiple competing providers offer broadly comparable capabilities on meaningful evaluations or common uses, not equality on every task.
- Similar products means general-purpose chat assistants or language-model APIs with overlapping uses.
- Sometimes permits cheaper offerings within competing product families; it does not require every rival's strongest model to cost less.
- Pricing comparisons use corresponding token charges or comparable access tiers, not company revenue or profit.
Counts as supported if: No substantial GPT-4 update occurred during the preceding year, while multiple rivals approached its capabilities and offered overlapping products, including materially cheaper options.
Counts as contradicted if: A substantial GPT-4 update occurred during the year, or credible comparisons reject the claimed competitive convergence or cheaper comparable offerings. Different results across these clauses warrant mixed.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Mixed
Part of the claim holds and part does not, and the record says which part is which.
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
The competitive-pressure portion is broadly supported: several providers offered overlapping products and competitive evaluations, and some rival tiers had lower token prices. The claim of no major GPT-4 update conflicts with GPT-4 Turbo's substantial functionality changes. This result depends on reading major as a substantial update rather than requiring a new model generation. Catch-up was uneven: hard-task evaluations still showed advantages for Turbo.
Limitations
- Major is an evaluator interpretation, not a threshold supplied by Marcus.
- Several capability comparisons are vendor-reported.
- Cheaper rival tiers were not identical in capability to their providers' strongest models.
- Benchmark results depend on tasks and judging methods.
Evidence (5 items)
Supports the claim: Anthropic,
on par with other leading models.
Why it matters here: The release documents overlapping multimodal chat and API products and reports strong Opus benchmark performance. Sonnet's published input and output token charges were below GPT-4 Turbo's corresponding charges in e1, although Opus was more expensive and Sonnet was a lower capability tier.
- Publisher
- Anthropic
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Supports the claim: Google,
including text and coding.
Why it matters here: Google's evaluation disclosure supports competitive convergence on selected text, coding and multimodal tasks. These are vendor results using specified evaluation methods, not proof of universal superiority or immediate availability of every model tier.
- Publisher
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Supports the claim: Mistral AI,
top-tier reasoning capabilities.
Why it matters here: The release reports competitive benchmarks, including a position behind GPT-4, and documents an available API and chat demonstrator. This supports the presence of another credible competitor without establishing parity across tasks.
- Publisher
- Mistral AI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Goes against the claim: OpenAI,
GPT‑4 Turbo is more capable
Why it matters here: The announcement documents an accessible GPT-4 Turbo preview with 128K context, improved instruction following, JSON mode and image-input support in the API. These constitute a substantial update under the stated reading of major.
- Publisher
- OpenAI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Context only: LMSYS Org,
the GPT-4 judge favoring GPT-4 outputs.
Why it matters here: The challenging-task benchmark retains a substantial GPT-4 Turbo lead under GPT judging, while the ordering and gaps change with Claude judging. It qualifies broad claims of parity and also shows why the original GPT-4 and later Turbo variants must be distinguished.
- Publisher
- LMSYS Org
- Kind of source
- Benchmark result
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, The Sam Altman Playbook
In the last year, his company has released no major update to GPT-4, and many competitors have caught up or are close. Competitors are eroding the initial gap in quality and starting to offer similar products, sometimes at significantly lower prices.
- Where
- paragraph 52
- Published
- 2024-05-05
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-the-sam-altman-playbook-6221c7-a1: Current. Outcome: Mixed. 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.