Quick answer: OpenAI has started a limited preview of GPT-5.6 Sol, Terra and Luna. During the preview, OpenAI says access is limited to selected trusted partners and organizations through the API and Codex; it is not available in ChatGPT yet, and there is no public waitlist.
If you run a business, agency, software team or AI workflow, the practical move is not to wait for hype. Use this short checklist to prepare your prompts, data boundaries, evaluation tests and cost controls before GPT-5.6 becomes more broadly available.
GPT-5.6 preview checklist for businesses
- Do not promise GPT-5.6 access to clients yet. OpenAI says preview access is limited and not self-service.
- Map use cases by model tier: Sol for hardest reasoning/security/science work, Terra for balanced daily workflows, Luna for high-volume lower-cost tasks.
- Create an evaluation set now: 20–50 real tasks from your business, with expected answers, pass/fail criteria and human review notes.
- Prepare cost limits: test token usage, caching assumptions and approval rules before putting GPT-5.6 into production.
- Review data and security policy: decide what data can go to an external model, who can use advanced cyber features, and what logs must be retained.
- Keep fallback models ready: do not break existing workflows while preview access is limited or rolling out gradually.
What OpenAI announced
OpenAI announced a limited preview of the GPT-5.6 model family on June 26, 2026. The three tiers are:
| Model | Positioning | Best early use case |
|---|---|---|
| GPT-5.6 Sol | Flagship, highest-capability model | Complex coding, deep reasoning, advanced analysis, security review |
| GPT-5.6 Terra | Balanced lower-cost option | Everyday business workflows, research, document tasks, agentic work |
| GPT-5.6 Luna | Fastest and most cost-efficient tier | High-volume support, summarization, classification and lightweight automation |
OpenAI says the family advances work in software engineering, computer use, professional knowledge work, scientific research and cybersecurity. The company also introduced new reasoning options, including a max reasoning effort for deeper work and an ultra mode that can use subagents for complex tasks.
Availability: who can access GPT-5.6 now?
According to OpenAI’s Help Center, GPT-5.6 is currently available only to a limited group of trusted partners and organizations through the OpenAI API and Codex. OpenAI says GPT-5.6 is not available in ChatGPT during the preview.
Important access notes:
- The preview is not a broad self-service program.
- There is no public application or waitlist.
- Access is scoped to approved API organizations and/or Codex workspaces.
- API approval does not automatically mean Codex approval, and vice versa.
- OpenAI says broader availability is planned in the coming weeks, but no exact general availability date has been announced.
GPT-5.6 pricing listed by OpenAI
OpenAI’s Help Center lists these preview prices per 1 million tokens:
| Model ID | Input / 1M tokens | Output / 1M tokens |
|---|---|---|
gpt-5.6-sol | $5.00 | $30.00 |
gpt-5.6-terra | $2.50 | $15.00 |
gpt-5.6-luna | $1.00 | $6.00 |
For teams planning production use, the key question is not only token price. Measure total task cost: retries, tool calls, long context, human review, caching and failed outputs.
Why the rollout is limited
OpenAI says it previewed the model capabilities and release plans with the U.S. government before launch and began with a limited partner preview at the government’s request. OpenAI also says it does not want this kind of government access process to become the long-term default.
Independent reporting from Axios and The Guardian also described the release as a staggered rollout connected to U.S. government review of advanced AI systems, especially around cybersecurity capability.
Practical implementation plan before general availability
1. Build a task benchmark
Collect real examples: sales emails, support tickets, code changes, spreadsheets, research briefs, SEO tasks, vulnerability triage notes or operations documents. For each task, define what a good answer looks like.
2. Separate low-risk and high-risk workflows
Use low-risk workflows first: drafting, summarization, classification, internal research and non-sensitive automation. Keep legal, medical, financial, hiring, security and customer-facing outputs under stricter human review.
3. Create a model-routing policy
Route expensive models only to tasks that justify them. For example: Luna for simple extraction, Terra for everyday agentic work, Sol for complex analysis or difficult coding/security tasks.
4. Test prompt caching and cost controls
OpenAI mentions more predictable prompt caching for GPT-5.6. Before production use, test repeated prompts, long system instructions and common context blocks to understand actual billing behavior.
5. Prepare governance for cyber features
Because OpenAI highlights improved cybersecurity capabilities, businesses should define allowed and disallowed uses, require authorization for security testing, and document review trails for sensitive work.
FAQ
Is GPT-5.6 available in ChatGPT?
No. OpenAI says GPT-5.6 is not available in ChatGPT during the limited preview.
Can anyone join the GPT-5.6 preview?
No. OpenAI says the preview is limited to selected trusted partners and organizations. There is no public application or waitlist.
When will GPT-5.6 be generally available?
OpenAI says it plans broader availability in the coming weeks, but it has not announced a specific general availability date.
Should businesses rebuild workflows around GPT-5.6 now?
Not yet. Prepare tests, routing rules and governance now, but keep production systems on currently available models until access, pricing and reliability are confirmed for your account.