The AI Read
← Predictions
tech · prediction T1

Anthropic keeps Model 2 unreleased through February 2027

Open·made August 16th 2026·resolves March 1st 2027
70% confidence

Why I called it

Anthropic disclosed a model stronger than its flagship and said it has no current plans to release it. Competitive pressure from OpenAI's IPO-season shipping cadence argues for releasing; the Risk Report's own reasoning, a saturated danger and rising uncertainty, argues against. Which force wins says how much the safety framework actually constrains the roadmap.

What countsWrong if the model the August Risk Report describes as Model 2, under that name or a successor product name Anthropic identifies with it, becomes available to paying or consumer customers before March 1st 2027. Internal use, safety research access and derivatives do not count.
What I based it on

The call, in full. The internal model Anthropic's August 2026 Risk Report calls Model 2 is not made generally available to paying customers before March 1st 2027.

Scoring criterion. RESOLVES WRONG if Anthropic makes Model 2 (or a product Anthropic publicly identifies as that internal model) generally available to paying customers before 2027-03-01, per Anthropic's own announcements. RESOLVES CORRECT otherwise.

The criterion is the machine-checkable version: a prediction that cannot be settled by a third party against a public source fails the build before it reaches this page.

Related