Why GPT-5.6 isn't usable yet (a US-requested limited rollout)
GPT-5.6: current availability
GPT-5.6 launched on June 26, 2026. But even if you want to try it, most people can't yet. It isn't about performance — OpenAI is deliberately widening access in stages.
Who can use it today (a developer-only limited release)
Right now, GPT-5.6 is a limited release available only through the developer routes — the API (a way for programs to call it) and Codex (OpenAI's coding assistant) — and only to a small group of trusted partners and organizations shared with the government. You can't reach it from the ChatGPT product most people use, so there's essentially no way for a general user to try it today.
"During the preview, GPT-5.6 models will initially be available through the API and Codex to a select group of trusted partners and organizations." — from the GPT-5.6 preview announcement
If you want the three models (Sol, Terra, Luna) — performance, pricing, and new features — see the GPT-5.6 Sol explainer. This article sticks to why it isn't usable, and when it will be.
Why it's being narrowed (the US request and staggered release)
The reason access is narrowed is a request from the US government. OpenAI says it is starting with a small group of trusted partners shared with the government, then widening from there — releasing a highly capable model carefully, a little at a time, rather than to everyone at once.
"At their request, we are starting with a limited preview for a small group of trusted partners whose participation has been shared with the government, before releasing more broadly." — from the GPT-5.6 preview announcement
Behind this is a new government rule. This month, the US set an executive order letting AI developers give the government up to 30 days to review their most capable models before they go out — so the government can check the most capable ones in advance, for security reasons.
"President Donald Trump signed an executive order earlier this month establishing a voluntary framework for AI developers to offer 'covered frontier models' to the U.S. government for up to 30 days before releasing them to trusted partners." — Reuters reporting
That said, OpenAI is clear that this government-mediated approach should not become the long-term norm. It sees broad, open access as the real goal and frames this limited release as temporary.
"We don't believe this kind of government access process should become the long-term default." — from the GPT-5.6 preview announcement
When it becomes usable (the general-release outlook)
If you wait, it becomes usable. OpenAI says it plans to make GPT-5.6 Sol, Terra, and Luna generally available — open to everyone — within the coming weeks, after more testing. Once that happens, you'll be able to use it from ChatGPT too.
"We believe in broad access, and we plan to make GPT-5.6 Sol, Terra, and Luna generally available in the coming weeks." — from the GPT-5.6 preview announcement
So GPT-5.6 being unusable right now isn't a fixed restriction — it's a queue ahead of a general release expected within weeks. If you're not in a hurry, waiting is the simplest move.
What it has in common with Anthropic's Fable 5 / Mythos 5 suspension
OpenAI vs. Anthropic: contrasting access control
GPT-5.6's limited rollout isn't a one-off. Around the same time, Anthropic's latest models also became unusable through government involvement. Putting the two side by side makes the bigger picture clear.
Fable 5 and Mythos 5 stopped under a US order
On June 12, 2026, Anthropic stopped its latest models — Fable 5 and Mythos 5 — worldwide, under a US export control directive (a government order that restricts exports on security grounds). The order called for cutting off access for every foreign national, inside or outside the United States. Anthropic's own foreign-national employees are included.
"The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees." — from the statement
For the full timeline and the alternatives you can use while it's down, see the Claude Fable 5 explainer.
The shared thread: security-driven access control
What the two cases share is this: in the name of national security, the government is controlling who can access the most capable AI. Neither was stopped for being weak. On the contrary — because the models are highly capable, and powerful even in areas like cyberattacks, who gets them and when became something to control. Capability itself is starting to be the reason for regulation.
The difference: narrowed "before" vs. stopped "after"
The picture is shared, but it plays out in opposite ways. OpenAI's is a "before" approach — it narrows who gets access ahead of release, and there's an exit in sight: a general release within weeks. Anthropic's was an "after" form: already public, then stopped on an order. Anthropic itself objects, arguing that pulling a commercial model used by hundreds of millions over one narrow jailbreak (a trick that slips past an AI's safety limits) is excessive.
"We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people." — from the statement
For users, GPT-5.6 feels like "usable in a little while," while Fable 5 feels like "it was working, then suddenly stopped." With Claude, only Fable 5 and Mythos 5 are affected; other models such as Opus 4.8 work as usual.
"Access to all other Anthropic models will not be affected." — from the statement
What you can do while you wait for GPT-5.6
Preparing for a model being cut off by regulation
Even if GPT-5.6 isn't usable right now, what to do is clear. Let's look at how to wait, and how to prepare for stops like these.
Get the work done on today's models while you wait
The general release is expected within weeks, so if you're not in a hurry, waiting is the safe choice. Almost everything until then can be handled by models you can already use. For writing, summarizing, or coding help, ChatGPT's GPT-5 line or Claude's Opus 4.8 work today. Unless you specifically need a GPT-5.6-only feature (such as ultra mode, which splits a job across several AI workers), your work won't stall.
For ChatGPT's model lineup, see the ChatGPT GPT-5 explainer; for an overview of Claude, see the What is Claude explainer.
Use more than one model to stay safe
These two cases show the risk of leaning entirely on one model. However capable it is, a model's release can be delayed or suddenly stopped by a regulatory or security decision. If you keep important work runnable on two or more models, work doesn't stop when one becomes unavailable. From the experience of using several AI models day to day, simply lining up an interchangeable backup per use case shrinks the impact of situations like these.
Official announcements are often long English documents, so reading the original is the surest way to track availability. Converting a web page into Markdown keeps the headings and lists intact, which also helps an AI read it and pull out the key points more accurately.
Summary: GPT-5.6 should be usable within weeks
GPT-5.6 isn't usable right now not because of performance, but because the US government asked for access to be widened gradually. For now it's open via the API and Codex to a small group of partners, and it isn't in ChatGPT yet — but a general release for everyone is expected within weeks. Set next to the parallel suspension of Anthropic's Fable 5 and Mythos 5, the same picture emerges: in the name of national security, access to the most capable AI is being controlled by the government. Get urgent work done on models you can use today, keep more than one model in rotation, and wait for the general release — that's the most practical stance right now.
