DEVELOPINGAIUpdated 39m ago

OpenAI unveils next-generation reasoning model, claims state-of-the-art on math and code

OpenAI announced a new reasoning-focused model with substantial gains on math, code, and scientific benchmarks. Independent evaluations are pending.

Source ConfidenceHIGH

14 sources analyzed · multi-source clustered

What happened

OpenAI introduced a new frontier model positioned as a successor to its reasoning line, citing improvements on competition math, coding tasks, and expert-level science questions. The company published benchmark numbers and a limited system card; pricing and general availability are staggered.

Key facts

·Model announced at a live event and via a company blog post.
·OpenAI reports gains on AIME, GPQA, and SWE-Bench-Verified.
·API rollout begins today for select tiers; consumer access follows.
·System card notes remaining limitations on long-horizon agents.

Why it matters

Reasoning models increasingly power agentic workflows across coding, research, and enterprise automation. Benchmark leadership shapes procurement decisions and downstream chip demand.

Confirmed

· OpenAI publicly announced the model.
· Blog post and system card are live on openai.com.
· API availability begins in staged rollout.

Remains unclear

· Independent third-party benchmark reproductions.
· Full pricing across all tiers and regions.
· Exact training compute and data provenance.

Differing reports

ReutersFrames the release as an incremental step over prior reasoning models.

The InformationReports internal debates about safety review timelines.

BloombergEmphasizes enterprise pricing implications.

Claims tracker

CONFIRMEDOpenAI announced the model publicly.

DISPUTEDModel outperforms all competitors on every benchmark.

UNCLEAREnterprise pricing has been finalized worldwide.

REPORTEDGeneral consumer availability begins today.

What happens next

→ Independent evaluators to publish reproductions within days.
→ Enterprise partners expected to announce integrations this week.
→ Regulatory attention on capability disclosures.

Timeline

9:00 AM
Livestream begins
OpenAI opens a public livestream.
9:14 AM
Model announced
Company confirms release and posts blog.
10:32 AM
API access opens
Staged rollout starts for select tiers.
1:05 PM
System card published
Limitations and eval details released.

How this story changed

INITIAL REPORT

Livestream teased; unnamed insiders leak details.

→

OFFICIAL CONFIRMATION

OpenAI publishes blog and system card.

→

NEW DETAILS

API rollout and enterprise pricing hints emerge.

→

CURRENT STATUS

Awaiting independent benchmark reproductions.

Compare coverage

Publisher	Headline	Emphasis	Time
OpenAI	Introducing our next-generation reasoning model	Primary announcement and system card	5h ago
Reuters	OpenAI launches new reasoning model	Wire coverage, incremental framing	5h ago
Bloomberg	OpenAI unveils reasoning model with enterprise push	Enterprise pricing and business impact	4h ago
The Information	Inside OpenAI's newest model launch	Internal review and timeline	4h ago
The Verge	OpenAI's new model tackles harder math	Benchmarks and demos	3h ago
AP	OpenAI debuts new AI model	Neutral summary	3h ago

Newsei does not automatically assign political labels to publishers.