OpenAI launched its newest model family on June 26 — but most enterprise customers cannot access it yet. GPT-5.6 Sol, Terra, and Luna are currently limited to a small group of government-vetted “trusted partners.” Hours later, the Trump administration partially lifted its block on Anthropic’s Mythos 5, clearing roughly 100 approved US organizations to access the model. Both moves signal a new default: frontier AI models with advanced cybersecurity capabilities now pass through a government vetting layer before reaching enterprise buyers.

Key Takeaways

• GPT-5.6 Sol, Terra, and Luna launched June 26 — limited to government-approved partners only with no GA date set.
• Anthropic Mythos 5 was cleared for ~100 US cyber defenders and infrastructure providers after a two-week full pulldown; Fable 5 (consumer model) remains unavailable.

What Did OpenAI Launch, and Why Can’t Enterprises Use It Yet?

GPT-5.6 Sol is OpenAI’s strongest model to date, designed for long-horizon agentic work with a new “ultra mode” for complex sub-agent tasks. Terra offers GPT-5.5-level performance at roughly half the cost; Luna is the fastest option.

The restriction on Sol stems from its advanced cybersecurity capabilities. AP News reports that the Office of the National Cyber Director and the Office of Science and Technology Policy requested a staggered launch, with Commerce Secretary Howard Lutnick separately pushing for even tighter controls — approving access “customer by customer” during the preview period.

OpenAI says Sol does not cross a “cyber critical threshold,” describing it as “better at helping people find and fix vulnerabilities than reliably carrying out end-to-end attacks.” The company is working with the administration on a repeatable review framework under EO 14409 and expects broader availability “in the coming weeks.”

Wired reports Commerce Secretary Lutnick cleared Mythos 5 on June 27 for approved US cyber defenders and infrastructure providers. The model had been fully pulled on June 12 after a government directive; Fable 5 remains offline while negotiations continue.

What Does a Government-Gated Model Launch Mean for Operators?

For most enterprises, the near-term impact is a timing gap: GPT-5.6 Sol is not available, with no firm GA date. OpenAI’s language — “coming weeks” — is intentionally vague.

The structural problem is larger than the wait. Operators building workflows or vendor evaluations around specific model capabilities now carry a new planning variable: government decisions can delay or pull a commercial model with limited notice. Anthropic’s Mythos experience — two weeks of full unavailability with no clear restoration timeline — is the clearest precedent. If your organization depends on AI-powered cybersecurity tooling from OpenAI or Anthropic, the underlying model version matters and could change.

The approval criteria for “trusted partner” status are not public. OpenAI has not explained how commercial enterprises can qualify or apply.

What to do now:

• Confirm with your OpenAI account team when GPT-5.6 reaches general availability for your contract.
• If your security vendor uses OpenAI or Anthropic models in a managed service, ask which model version is running and what the continuity plan is if access is restricted.
• Add model-access continuity and government-restriction scenarios to AI vendor due diligence — especially for regulated industries.

What to watch next: The August 1 EO 14409 framework deadline is the key signal. Published approval criteria would make this process predictable; without them, ad hoc restriction-and-restoration cycles remain the default. Also watch whether Terra and Luna receive GA clearance before Sol.

---

Frequently Asked Questions

Does a ChatGPT Enterprise contract give me access to GPT-5.6 Sol now?

No. As of June 28, GPT-5.6 Sol is available only to organizations approved through the government vetting process. Standard Enterprise accounts are not automatically included. OpenAI expects broader availability “in the coming weeks” but has not given a specific date. Ask your account team for your contract-specific timeline.

Who can access Anthropic Mythos 5 now?

As of June 27, Mythos 5 is cleared for roughly 100 US cyber defenders and infrastructure providers per Commerce Secretary Lutnick’s approval. It is not available to general enterprise customers. Fable 5 (the consumer-facing model) remains offline while Anthropic and the White House continue negotiations.