On July 9, 2026, OpenAI opens GPT-5.6 to the general public. For several weeks the new model family had been limited to a short list of government-vetted partners. For an SME leader, the question is not "which model is the most impressive" but "what actually becomes available, at what price, and with what reliability guarantees." This article answers those three questions, with sources.

Key takeaways

• GPT-5.6 becomes publicly available on July 9, 2026, in ChatGPT and via the API, after weeks of restricted access (sources: Axios, franceinfo, Clubic).
• The family has three models: Sol (flagship), Terra (mid-tier, half the price), Luna (fastest and cheapest).
• For the first time, the White House shaped the release schedule of an American model, based on a June 2, 2026 executive order (sources: Qz, The Next Web).
• On July 8, the White House pushed back on having given a "formal" green light, while letting the launch proceed (sources: Yellow, Let's Data Science).
• For an SME, the practical stake is twofold: cheaper models for high-volume tasks, and a reminder that an AI tool's availability can now hinge on a regulatory review.

What launches on July 9, 2026

GPT-5.6 is not a single model but a three-tier lineup, designed so each use case pays a fair price: a powerful, expensive model for complex tasks, and a fast, cheap model for volume.

Source: Clubic, based on OpenAI's announced API pricing. Sam Altman confirmed the launch on X: "GPT-5.6 sol launches thursday."

On published benchmarks, Sol scores 88.8% on Terminal-Bench 2.1, a test that measures a model's ability to run command-line tasks, versus 88% for a leading rival model (source: Clubic). The gap is small: the real differentiator for an SME will be less the raw score than the capability-to-price ratio of Terra and Luna.

Why the launch was delayed

This is the genuinely new part of the story, and it goes beyond a model release. An executive order signed on June 2, 2026, titled "Promoting Advanced Artificial Intelligence Innovation and Security," set up a process by which frontier model developers can give the government access to their most powerful models for up to 30 days before public release (sources: The Next Web, Business Standard).

Under it, the administration asked OpenAI to first limit GPT-5.6 to a small group of approved partners. It was the first time an American lab placed a frontier model behind a state-approved roster. The evaluation was run by the Commerce Department's Center for AI Standards and Innovation (CAISI), and OpenAI sent engineers to Washington to answer questions (sources: Qz, BankInfoSecurity).

One point deserves precision: on July 8, 2026, the White House denied having formally authorized the broadened access, even as several outlets announced availability on July 9 (sources: Yellow, Let's Data Science). In other words, the timing held, but the exact account of "who approved what" remains disputed. For a leader, the substance is what matters: the availability of a frontier model can now hinge on a regulatory step, which was unthinkable a year ago.

What it actually changes for an SME

Three practical consequences, without hype.

The underlying point stays positive and measured: each model iteration lowers the price of everyday tasks and brings within SME reach capabilities once reserved for large groups. Luna at $1 per million input tokens makes automations viable that were too costly two years ago. That steady price decline, more than the headline score, is what matters for your budget.

FAQ

Is GPT-5.6 available to everyone right away?

OpenAI announced public availability on July 9, 2026, in ChatGPT and via the API, with the usual usage limits. Rollout may be gradual by plan and region. Check availability in your account before planning a migration.

What is the difference between Sol, Terra and Luna?

Sol is the flagship model for complex tasks. Terra targets enterprise workloads at a reduced price (about half of Sol). Luna is the fastest and cheapest, built for high-volume tasks. The choice depends on your use case and budget, not on headline "power."

Why did the US government step in?

A June 2, 2026 executive order allows a government review of frontier models up to 30 days before release. GPT-5.6 was evaluated by the Commerce Department's CAISI, which explains the weeks of restricted access before the public launch.

Is a European SME affected by this US review?

The review is a US procedure. A European SME is not directly subject to it, but feels the indirect effect: the release date of a tool it uses can depend on it. That is one more reason not to build a critical process on a single model.

Conclusion

The public launch of GPT-5.6 marks two trends: models increasingly segmented by price, and availability now tied to regulatory steps. For an SME, the right posture is neither hype nor wait-and-see: test the new tiers on your real tasks, pay the right price, and keep an alternative. To go further, compare models by use case in our Mag resources or see how other SMEs approach it in our success stories.