OpenAI gets US government clearance for broad GPT-5.6 launch

In July 2026, the US government gave the green light for the broad release of GPT-5.6, the most powerful model OpenAI has deployed. For several weeks the model stayed in a closed preview with about 20 partners while Washington carried out a government assessment under a new oversight regime for frontier AI systems.

Why was GPT-5.6 delayed at launch?

The delay stems from a new regulatory mechanism in the US that covers the most powerful AI models, those classed as "frontier". The regime requires developers to undergo a government assessment and obtain official approval before releasing a product widely. By design, it is the equivalent of the certification process in aviation or pharmaceuticals, but for generative AI.

GPT-5.6 spent several weeks in this review. Throughout that time the model was available only to a limited group: about 20 organizations received closed preview access. The partners' names have not been disclosed.

• GPT-5.6 is the most advanced model OpenAI has deployed as of July 2026
• Limited-access period: several weeks of government review
• Organizations in the closed preview: about 20
• Legal mechanism: the new US oversight regime for frontier AI systems
• Outcome: approval received, broad launch unblocked

A precedent for OpenAI's competitors

GPT-5.6 is the first model to go publicly through the new US oversight regime, from limited preview to broad launch. It sets a precedent for the whole industry: it is now clear that the procedure takes several weeks and includes a government review involving a closed group of testers.

For Anthropic, Google DeepMind, xAI and other companies building frontier systems, this means their next major release will follow the same path. The time between "the model is ready" and "the model is available to everyone" now includes a mandatory regulatory pause, which for GPT-5.6 lasted several weeks.

The details of the process itself, including the evaluation criteria, the list of mandatory tests and the makeup of the expert panel, remain opaque. GPT-5.6's passage through it creates the first public benchmarks.

What users will get

With the approval in hand, OpenAI is entitled to open GPT-5.6 to its whole audience through ChatGPT and the API. Functionality that ordinary users could not access during this period is becoming generally available.

Specific terms, including pricing plans, regional restrictions and the exact timeline for the phased rollout, are not specified in the source. The regulatory barrier is gone; the rest is up to OpenAI.

What this means

GPT-5.6's approval is the first public confirmation that the new US oversight regime for AI works: the model went through it and reached the market. For OpenAI, it is a win. For the industry, it is a signal that the era of instant releases without government involvement is over. Every future flagship will now pass through a government sieve before it reaches users.