OpenAI has officially announced the start of global preview access to its new flagship artificial intelligence lineup — GPT-5.6. The public release of three model versions, named Sol, Luna, and Terra, will take place on July 9. The launch became possible only after intense negotiations with the US government and the resolution of all cybersecurity concerns.

Delay Due to Trump’s Executive Order

Initially, the company planned to deploy GPT-5.6 by the end of June. However, plans were disrupted by a new executive order from President Donald Trump concerning cybersecurity in the AI sector. The document requires developers to voluntarily submit their powerful models for government review at least 30 days before public launch.

As a result, initial access to GPT-5.6 was granted only to a few trusted US government agencies. OpenAI publicly stated that such total control should not become standard practice, but agreed to cooperate to avoid delaying the product’s market entry.

Technical Diplomacy in Washington

To accelerate the process, the company sent a team of leading technical engineers directly to Washington. There, together with specialists from the AI Standards and Innovation Center at the US Department of Commerce, they conducted additional stress tests. As a result, all government concerns regarding model security were promptly addressed, and convincing the White House took less than a month.

Three Models for Different Tasks

The new GPT-5.6 lineup is optimized for various tasks and user budgets. OpenAI offers three options:

  • Sol — the most powerful and uncompromising neural network in the company’s history. It is designed for complex computations and large-scale corporate tasks. Usage costs $5 per million input tokens and $30 per million output tokens.
  • Terra — a balanced model for everyday use. Developers promise its performance matches the previous flagship version, GPT-5.5. At the same time, Terra operates at half the cost: $2.5 per million input tokens and $15 per million output tokens.
  • Luna — an ultra-budget and most accessible option, intended for fast and mass operations. The price is just $1 per million input tokens and $6 per million output tokens.

Impact on Competitors

OpenAI is not the only company affected by the strict new White House policy. Its main competitor — startup Anthropic — was previously forced to completely block access to its advanced cybersecurity model Mythos and consumer model Fable. The government issued a strict requirement to fully exclude the possibility of these technologies being used by foreign nationals.

After undergoing similar reviews, Anthropic also received the green light to deploy its exclusive Mythos 5 model. In the near future, the company plans to bring its public version, Fable 5, back to the market.