GPT-5.6 (Sol, Terra, Luna)
OpenAI's next-generation family, previewed 26 June 2026 — three models, ~20 partner organizations, and a US government safety review standing between it and general availability.
GPT-5.6: what's confirmed so far
| Maker | OpenAI |
| Preview began | 26 June 2026 |
| Status | Limited preview — ~20 partner organizations, government safety review |
| Broad availability | Promised "in the coming weeks" (not yet dated) |
| Variants | Sol (flagship) · Terra (balanced) · Luna (fast/affordable) |
| API pricing | TBD — not published during preview |
| Context window | TBD — not published during preview |
| Benchmarks | TBD — no public model card yet beyond the preview system card |
Calculate LLM API costs → This model has no confirmed USD per-token rate card, so it is not in the calculator — compare 16 published rate cards instead.
The three models: Sol, Terra, Luna
Sol is the flagship — OpenAI's next frontier model, launching with what the company calls its most robust safety stack to date, including strengthened protections for higher-risk activity, sensitive cyber requests and repeated misuse.
Terra is the balanced everyday-work model. OpenAI reports it delivers performance competitive with GPT-5.5 while being 2x cheaper — which, if the claim holds at GA, would be the more commercially significant release of the two.
Luna is the fast tier: strong capability at OpenAI's lowest cost, aimed at the same high-volume segment as Claude Haiku 4.5 and Gemini Flash models.
Why it's restricted
Unusually, GPT-5.6 shipped first to a limited group of roughly 20 trusted partner organizations at the request of the US government, gated behind a government safety review. This mirrors the pattern of mid-2026 frontier releases — Claude Fable 5 was briefly pulled under export controls the same month — and marks a new level of state involvement in frontier model deployment.
Should you wait for GPT-5.6?
- If you're on GPT-5.5 today — nothing to do yet; the API surface for general customers hasn't opened. Terra's claimed 2x price-performance is the number to watch.
- If you're choosing a stack right now — decide on what's shipping: GPT-5.5, Claude Opus 4.8 and Gemini 3.5 Flash are all generally available.
How GPT-5.6 compares
- GPT-5.5 ("Spud") — the current shipping OpenAI flagship →
- Claude Fable 5 — the other government-entangled frontier release →
- Gemini 3.5 Flash — Google's newest production model →
Read OpenAI's preview announcement
Frequently asked questions
Is GPT-5.6 released?
Not broadly. OpenAI began a limited preview on 26 June 2026 with roughly 20 trusted partner organizations, gated behind a US government safety review. Broad availability was promised "in the coming weeks."
What models are in the family?
Sol (flagship), Terra (balanced — ~GPT-5.5 performance at 2x cheaper) and Luna (fast, OpenAI's lowest cost).
How much does GPT-5.6 cost?
Public API pricing has not been announced during the preview. Exact per-token rates are TBD.
Why is it restricted?
At the request of the US government, OpenAI shipped it to a limited partner group behind a government safety review. Sol carries OpenAI's most robust safety stack to date.
Is GPT-5.6 the same as GPT-6?
No. GPT-6 has not been released; GPT-5.6 is an intermediate family between GPT-5.5 and GPT-6.