“Playing the field with Mistral wasn’t enough; now he is building a really big model, big enough to compete directly with GPT-4, internally:”
Our reading, the sentence that was checked:
Beyond partnering with Mistral, Microsoft is building a large model in-house that is big enough to compete directly with GPT-4.
Assessment: Supported Evidence as of
Where and when Gary Marcus wrote it
Marcus on AI, Microsoft and OpenAI’s increasingly complicated relationship, paragraph 13.
- Said
- Source observed
- live page, full text, public
- Evidence as of
- the cutoff for this assessment
- Assessment recorded
- assessment marcus-microsoft-and-openais-increasingly-d040d5-a1
We recorded 5 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 2024-05-06. 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
- per The Information's reporting (qualifier, stated by the speaker)
From the speaker
- Playing the field with Mistral
- now he is building
- a really big model
- big enough to compete directly with GPT-4
- internally
- The Information reports
Our interpretation
- The claim concerns an internal development project reported on the publication date, so the target is the publicly reportable state on May 6, 2024.
- Big enough to compete means a project positioned as a potential GPT-4-class competitor, not a released model with demonstrated benchmark parity.
- No numerical size or performance threshold is supplied.
- The attribution to The Information qualifies the evidence; it is not a conditional forecast.
Counts as supported if: Reliable contemporaneous reporting identifies an internal Microsoft large-model project positioned to compete with GPT-4, with Microsoft's Mistral relationship established separately.
Counts as contradicted if: Reliable records establish that no such project existed, that it was solely an externally developed model, or that it was not intended to compete in GPT-4's class.
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 May 6 reporting describes MAI-1 as a large internal Microsoft project intended to compete with leading models, including GPT-4. Mistral's earlier announcement confirms the partnership mentioned in the lead-in. This supports a development-and-positioning claim. It does not establish that MAI-1 had already matched GPT-4: the reports made its eventual use dependent on performance.
Limitations
- Support depends on reading compete as intended competitive positioning.
- The detailed project reporting ultimately relies on The Information.
- Reported parameter counts do not demonstrate equal capabilities.
Evidence (3 items)
Supports the claim: The Decoder,
compete with GPT-4
Why it matters here: The original article identifies MAI-1 as Microsoft's internal project and describes its intended competitive class. Its underlying source is The Information, rather than an independent technical evaluation.
- Publisher
- The Decoder
- Kind of source
- Secondary summary
- Describes events of
- Public from
Supports the claim: Reuters via Yahoo Finance,
The new model
Why it matters here: Reuters describes MAI-1 as an in-house project and reports a planned scale of roughly 500 billion parameters. It also says the intended use depended on performance and that Microsoft declined comment, limiting any inference of proven parity.
- Publisher
- Reuters via Yahoo Finance
- Kind of source
- News report
- Describes events of
- Public from
Context only: Mistral AI,
first distribution partner
Why it matters here: Mistral's announcement establishes the earlier partnership referenced by Marcus. Its February event date makes it background to the May 6 development observation.
- Publisher
- Mistral AI
- Kind of source
- Company disclosure: the company describing itself, not an independent finding
- Describes events of
- Public from
Where it was said (1)
Said once in the sources this register covers.
: Marcus on AI, Microsoft and OpenAI’s increasingly complicated relationship
Playing the field with Mistral wasn’t enough; now he is building a really big model, big enough to compete directly with GPT-4, internally:
- Where
- paragraph 13
- Published
- 2024-05-06
- Source observed
- , live page
- Access
- full text, public
History
Assessment versions
- Assessment marcus-microsoft-and-openais-increasingly-d040d5-a1: 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.
About assessment marcus-microsoft-and-openais-increasingly-d040d5-a1.