Breaking

OpenAI to Stagger GPT-5.6 Launch After US Review Request

June 25, 2026 at 16:34 EDT

Foundation Models
Policy & Regulation

OpenAI plans to release its next model, GPT-5.6, in stages after the US Trump administration requested a pre-release review. The model will first be offered as a limited preview to a small group of trusted partners, then move to a broad public launch weeks later following government approval.

June 25, 2026 · OpenAI

Washington Asks OpenAI to Stagger Its Next Flagship Release

In a memo to staff, Sam Altman said GPT-5.6 would first ship as a limited preview — with the government approving access "customer by customer" — calling the approach "highly unusual" and "not our preferred long term model."

80–89%

Prediction-market odds of a GPT-5.6 release by end of June

Back-to-back labs facing government intervention: OpenAI & Anthropic

1.5M

Rumored GPT-5.6 context window, in tokens

Context window: current vs. rumored

Tokens the model can hold at once — built from 0.5M blocks

GPT-5.5 (current)

1.5M

GPT-5.6 (rumored)

The new release flow

Limited preview

small set of partners

→

Gov't review

approved customer by customer

→

Wider release

"sustainable process" TBD

The new normal

The era of "drop it and see what happens" is ending — regulators are at the table whether labs like it or not.

Commerce Secretary reportedly warned OpenAI not to launch without sign-off from other agencies.
Anthropic temporarily disabled its Fable 5 and Mythos 5 models over export-control and national-security concerns.
GPT-5.6 specs remain unconfirmed pending an official OpenAI announcement.

Continue reading

The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.