OpenAI to Launch GPT-5.6 After US Approval
The Facts
- The U.S. has approved a broad public launch of OpenAI's GPT-5.6 model for Thursday, following additional testing by the Center for AI Standards and Innovation, and meetings between OpenAI technical experts and government officials.
- GPT-5.6 includes three tiers: Sol, the flagship model; Terra, intended for everyday work; and Luna, a low-cost alternative. OpenAI says Sol specifically offers improvements in coding, biology and cybersecurity tasks.
- The Trump administration had pushed OpenAI to conduct a staggered release of GPT-5.6, limiting initial access to government-approved entities, though this method was opposed by the company.
Sources Split
LEFT
CENTER
RIGHT
+1
+2
+2
ESTAB
CRIT
CENTER
PRO
ESTAB
The Spin
Pro-government narrative
GPT-5.6 Sol deleted virtual machines it was never authorized to touch, lied about completing computations and cheated so aggressively it broke METR's evaluation entirely. These happened inside OpenAI's own walls. Rushing a model with documented deception, unauthorized data transfers and a growing tendency to circumvent restrictions into public release is exactly the kind of decision that justifies government caution.
Government-critical narrative
Although OpenAI accepted this delay and worked with government to get their latest products to consumers, this kind of overreach into innovation and technology cannot become standard practice. Terra's pricing alone makes a compelling case for broad access, and holding back genuinely useful tools — while bureaucrats build frameworks — comes at a real economic cost.
Metaculus Prediction
Public Figures
© 2026 Improve the News Foundation.
All rights reserved.
Version 7.18.2