“That wasn’t tested (presumably they don’t have access), and only a handful of benchmarks were actually reported.”
Our reading, the sentence that was checked:
The Grok 3 launch did not test against full o3 and reported only a handful of benchmarks.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Grok 3 Hot Take, paragraph 6.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-grok-3-hot-take-302078-a2
We recorded 9 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 2025-02-18. 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)
From the speaker
- the full but unreleased version of o3
- That wasn’t tested
- presumably they don’t have access
- only a handful of benchmarks were actually reported
Our interpretation
- The assertion concerns comparisons presented at the launch, rather than undisclosed internal experiments.
- Full o3 is distinct from o3-mini, including o3-mini at high reasoning effort.
- Handful means a small set of benchmark families. Model variants, repeated runs and leaderboard subcategories are not separate benchmarks.
- The access explanation is explicitly speculative and is not treated as a firm factual commitment.
Counts as supported if: The launch record excludes full o3 from its comparisons and reports a small set of benchmark families.
Counts as contradicted if: The launch includes a comparison with full o3, or reports a clearly extensive collection of distinct benchmark families.
Retrospective: these criteria were written after the outcome was public.
Current assessment
Was it true for the period described?
Outcome: Supported
The dated evidence matches what the claim committed to, under the criteria frozen before grading.
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 available launch evidence supports both parts of the claim. A February 18, 2025 observation identifies full o3's omission from the livestream chart [e1]. DeepLearning.AI's February 19 report identifies three core benchmark families—AIME 2024, GPQA and LiveCodeBench—and lists o3-mini at high effort among the reasoning-model comparisons [e2]. Together, these support a small set of reported benchmark families and no presented comparison with full o3 [e1, e2].
Limitations
- The complete primary launch charts were not readable, so the evidence does not establish an exhaustive benchmark count.
- “Handful” has no exact numerical cutoff.
- The public comparisons do not establish whether xAI privately tested full o3 or had access to it.
Evidence (2 items)
Supports the claim: Rex Asabor, OpenAI employee,
they omitted o3
Why it matters here: This contemporaneous observation explicitly identifies the omission of full o3. It establishes the public chart's scope, not whether xAI privately tested o3 or had access.
- Where
- Main post
- Publisher
- Rex Asabor, OpenAI employee
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Supports the claim: DeepLearning.AI,
math (AIME 2024)
Why it matters here: The report identifies three core benchmark families in the launch results. Its detailed competitor lists name o3-mini, rather than full o3, corroborating the limited scope of the comparisons.
- Where
- Results section
- Publisher
- DeepLearning.AI
- Kind of source
- Secondary summary
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Grok 3 Hot Take
That wasn’t tested (presumably they don’t have access), and only a handful of benchmarks were actually reported.
- Where
- paragraph 6
- Published
- 2025-02-18
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-grok-3-hot-take-302078-a2: Current. Outcome: Supported. 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.