OpenAI’s July 9, 2026 GPT-5.6 launch matters because it does more than introduce one stronger flagship model. It turns the model lineup into a clearer operating system for teams: one family, three durable tiers, multiple effort settings, and a more explicit split between everyday work and high-stakes work.
That is what Sol, Terra, and Luna are really about. They are not just poetic names. They are OpenAI’s way of telling teams to stop thinking in terms of one default model for everything and start matching intelligence, speed, and cost to the workflow in front of them.
AI Search Snapshot
On July 9, 2026, OpenAI launched GPT-5.6 for general availability and said the family includes Sol as the flagship tier, Terra as the balanced everyday-work tier, and Luna as the fastest and most affordable tier. OpenAI also tied GPT-5.6 directly to ChatGPT Work, Codex, and the API, while adding higher-effort settings such as max and ultra for more demanding work.
Direct Answer
GPT-5.6 is OpenAI’s new model family for production work across ChatGPT, ChatGPT Work, Codex, and the API. The key idea is simple: Sol is for the hardest work, Terra is for the broad middle, and Luna is for speed and affordability at scale. OpenAI is also making effort level part of the product choice, not only the model choice, which means teams now have to pick both a tier and how much reasoning time to spend.
For business users, the launch matters because it lines up with the new GPT-5.6 hub story inside ChatGPT Work: a system that can take action across files and apps, stay with projects for hours, and turn goals into finished work. For technical teams, it matters because the API now exposes the same three-tier family with programmatic tool calling and multi-agent options.
What OpenAI Announced on July 9, 2026
| Release point | What OpenAI said | Why it matters | Best first question to ask |
|---|---|---|---|
| Model family | GPT-5.6 launched as Sol, Terra, and Luna. | The lineup is now structured around durable capability tiers, not only one monolithic flagship. | Which tasks actually need flagship intelligence? |
| Reasoning settings | OpenAI introduced max and expanded ultra positioning. |
Effort level now changes the product decision, not just the prompt. | When is extra reasoning time worth the cost? |
| Surface area | GPT-5.6 is available across ChatGPT, ChatGPT Work, Codex, and the OpenAI API. | The same model family can now power end-user work and developer workflows. | Do we need a user-facing tool, an API build, or both? |
| Distribution | OpenAI also said GPT-5.6 becomes the preferred model in Microsoft 365 Copilot. | The model family is being pushed into mainstream knowledge-work tools quickly. | Where will users encounter GPT-5.6 first? |
| Pricing | OpenAI published separate token prices for Sol, Terra, and Luna. | Model choice is now an operating-leverage decision as much as a quality decision. | Where can we shift routine work to a lower-cost tier? |
Sol vs Terra vs Luna in Plain English
The easiest mistake is to read the names as a simple ranking from best to worst. That is not how OpenAI frames them. OpenAI says Sol is the flagship, Terra is balanced for everyday work, and Luna is fastest and most affordable. In practice, that means each tier is supposed to win a different job.
| Tier | How OpenAI positions it | Best-fit workflows | Main tradeoff |
|---|---|---|---|
| Sol | Flagship model for the hardest work. | Complex analysis, deeper coding, high-stakes document work, long-running agent tasks. | Higher cost than Terra or Luna. |
| Terra | Balanced model for everyday work. | Most business tasks, ongoing team workflows, draft creation, analysis that must stay strong but not maximal. | Not the best choice when the hardest reasoning quality matters most. |
| Luna | Fastest and most affordable model. | Large-volume tasks, lightweight transformations, lower-cost automation, quick iterations. | Lower ceiling than Sol on difficult work. |
If you are already familiar with our OpenAI API explainer, this is the same mindset we recommend there: do not pay for frontier reasoning when the workflow really needs throughput, and do not optimize for cheap tokens if the work is expensive to get wrong.
What Max and Ultra Actually Change
GPT-5.6 is also important because OpenAI is making effort settings explicit. According to the official launch and preview materials, max gives GPT-5.6 more time to reason, check alternatives, and revise. ultra goes further by coordinating multiple agents in parallel for demanding tasks.
That changes how teams should evaluate the release. Before, many people asked only, “Which model should we use?” Now the better question is, “Which model and which effort setting should we use for this job?” A Terra or Sol workflow at normal effort can behave very differently from the same workflow at max or ultra.
For most teams, max is easier to justify than ultra. It fits work where the answer quality or plan quality matters enough to wait longer. ultra is more for jobs that benefit from parallel decomposition, such as larger research, synthesis, or multi-step technical work.
Where GPT-5.6 Is Available Today
OpenAI’s own release page makes the rollout structure clearer than many headlines do. In Chat, Plus, Pro, Business, and Enterprise users access Sol through higher-effort settings, while Pro and Enterprise can also select Sol Pro for especially complex tasks. In ChatGPT Work and Codex, Free and Go users access Terra, and paid users can choose among Sol, Terra, and Luna. In the API, developers can access all three tiers directly.
This matters because GPT-5.6 is not only an API story. OpenAI is clearly trying to line up the same family across consumer-style chat, work-oriented agents, developer tooling, and enterprise productivity surfaces. The Microsoft 365 Copilot announcement reinforces that point: GPT-5.6 is being framed as useful for Word, Excel, PowerPoint, Chat, and Cowork workflows, not only standalone AI experiments.
Why This Matters for ChatGPT Work, Codex, and the API
The ChatGPT Work announcement is what makes the GPT-5.6 release feel bigger than a spec sheet refresh. OpenAI says ChatGPT Work can gather information across apps and workflows, create finished materials like slides and docs, and stay with complex projects for hours. GPT-5.6 is the model family underneath that experience.
That means the model tiers are no longer abstract. In ChatGPT Work, the choice between Sol, Terra, and Luna can affect whether a workflow feels like a thoughtful operator, a practical day-to-day assistant, or a fast utility layer. In Codex, the same choice affects how much time and cost you want to spend on planning, editing, and tool coordination. In the API, it affects whether your system is optimized for accuracy, balance, or scale.
For teams already using ChatGPT Projects or recurring task flows like the ones in our ChatGPT Tasks guide, GPT-5.6 creates a more intentional path: keep routine work cheap where possible, but step up to Sol and higher effort when the work product really matters.
What Teams Should Not Assume Yet
- Do not assume Sol is automatically the right default for every workflow. Many routine internal jobs will probably fit Terra better.
- Do not assume cheaper always means better operations. Luna is attractive for throughput, but expensive mistakes can wipe out token savings.
- Do not assume
ultrais the new default. It is a strong option for hard work, not a reason to over-invest on basic tasks. - Do not assume a surface like Microsoft 365 Copilot gives you the same level of control as direct API usage.
- Do not assume the family structure replaces workflow design. Better models still need better task boundaries.
FAQ
What is GPT-5.6 in one sentence?
It is OpenAI’s July 9, 2026 model family built around Sol, Terra, and Luna, with explicit effort settings for different kinds of work.
What is the difference between Sol, Terra, and Luna?
Sol is the flagship tier, Terra is the balanced everyday-work tier, and Luna is the fastest and most affordable tier.
Does GPT-5.6 power ChatGPT Work?
Yes. OpenAI says ChatGPT Work is powered by GPT-5.6 and is designed for longer, more involved multi-step work.
Is GPT-5.6 only for developers?
No. OpenAI launched it across ChatGPT, ChatGPT Work, Codex, the API, and Microsoft 365 Copilot-related experiences.
What should I read next?
Start with our Sol vs Terra vs Luna choosing guide if you need a practical decision rule, or the small business program explainer if you want to understand the most immediate business rollout angle.
Bottom Line
GPT-5.6 matters because OpenAI is not just shipping a stronger model. It is teaching users to think in tiers, effort levels, and workflow fit. Sol, Terra, and Luna are really operating choices about where to spend intelligence, time, and money across ChatGPT Work, Codex, and the API.
Verified External Sources
- OpenAI: GPT-5.6: Frontier intelligence that scales with your ambition
- OpenAI: Previewing GPT-5.6 Sol: a next-generation model
- OpenAI: ChatGPT is now a partner for your most ambitious work
- OpenAI: GPT-5.6 is now the preferred model in Microsoft 365 Copilot