OpenAI restricts GPT-5.6 rollout after US government request, warns against permanent controls

OpenAI announced Friday that it is limiting the initial release of its next-generation GPT-5.6 family of AI models to a small group of trusted partners, following a direct request from the Trump administration. The company made clear it does not view this arrangement as a sustainable long-term policy.

Three models, one restriction

The GPT-5.6 lineup includes three distinct models: Sol, the flagship model with advanced agentic capabilities; Terra, a balanced model for everyday use; and Luna, a faster, lower-cost option. Despite their different purposes, the U.S. government has restricted the release of all three. OpenAI stated that the preview is limited to partners whose participation has been shared with the government in advance.

The move follows a broader pattern of increased government scrutiny on frontier AI systems. Earlier this month, Anthropic released its powerful Fable 5 model, only to be ordered by the administration to remove access for foreign nationals. Anthropic ultimately took the model down entirely.

De facto licensing regime

Dean Ball, a former White House AI advisor who is soon to join OpenAI, argues that President Trump’s recent executive order has created a de facto involuntary licensing regime for frontier AI. The order asks certain AI companies to voluntarily submit their most advanced models for government review up to 30 days before release. In practice, Ball contends, this has led to heavy-handed restrictions without clearly defined safety standards.

Ball warns that the lack of transparent criteria could lead to endless launch delays, potentially handing an advantage to China in the global AI race and jeopardizing the billions of dollars being invested in AI infrastructure.

OpenAI pushes back

While OpenAI complied with the administration’s request, the company was explicit about its objections. We don’t believe this kind of government access process should become the long-term default, the company wrote in a Friday blog post. It keeps the best tools from users, developers, enterprises, cyber defenders, and global partners who need them.

OpenAI characterized the preview as a short-term step that will put GPT-5.6 on the path to broader availability in the coming weeks. The company says it is working with the administration to develop a new executive order framework on cybersecurity, as well as a repeatable process for future model releases.

GPT-5.6 Sol: Capabilities and safety

OpenAI describes GPT-5.6 Sol as its strongest model yet, with improved agentic capabilities in coding, biology, and cybersecurity. It introduces a max reasoning effort mode and an ultra mode that uses coordinated subagents to solve highly complex tasks. According to OpenAI, Sol is slightly better at coding workflows than Anthropic’s Claude Mythos 5, which the administration also effectively banned this month. Sol is also competitive with Mythos preview while using a third of the output tokens.

To address safety concerns, OpenAI says Sol includes its most robust security stack yet, heavily hardened against adversarial attacks and intentionally optimized to favor defensive cybersecurity work over offensive exploits. The company states that safety guardrails are built directly into the core model’s behavior, rather than relying on a separate filter on top of it.

This architectural choice appears designed to avoid the problems that plagued Anthropic’s Fable 5, where the model would silently route high-risk queries to an older model, causing user backlash and false positives.

Pricing and availability

GPT-5.6 comes in three sizes with tiered pricing: Sol costs $5 per million input tokens and $30 per million output tokens; Terra costs half that; and Luna costs $1 and $6, respectively. OpenAI says it has also improved prompt caching to make repeated prompts cheaper and more predictable.

While initially limited to a select group of partners, OpenAI plans to make the models more broadly available to users of ChatGPT, Codex, and the API in the near future.

Conclusion

The GPT-5.6 rollout marks a significant moment in the evolving relationship between AI companies and the U.S. government. OpenAI’s compliance, coupled with its public objection, signals that the industry is willing to negotiate but wary of permanent controls. The coming weeks will reveal whether a repeatable framework can be established, or whether the de facto licensing regime becomes the new normal for frontier AI development.

FAQs

Q1: Why is OpenAI limiting the release of GPT-5.6?
OpenAI is limiting the release at the request of the Trump administration, which asked the company to restrict access to a small group of trusted partners whose participation has been shared with the government.

Q2: What are the three models in the GPT-5.6 family?
The lineup includes Sol (the flagship model with advanced agentic capabilities), Terra (a balanced model for everyday use), and Luna (a faster, lower-cost option). All three are subject to the same restrictions.

Q3: When will GPT-5.6 be widely available?
OpenAI says the preview is a short-term step and that broader availability is expected in the coming weeks, as the company works with the administration to establish a repeatable process for future model releases.