GPT-5.6 is here: Sol, Terra and Luna. Which one do you actually need?

What is GPT-5.6?

GPT-5.6 is OpenAI's latest model generation, released publicly on 9 July 2026. It comes in three tiers: Sol (the flagship), Terra (balanced price and performance) and Luna (fastest and cheapest). GPT-5.6 is also the engine behind the new ChatGPT Work.

The number now tells you which generation a model belongs to, while Sol, Terra and Luna are fixed tiers that can be updated at their own pace. It sounds like a detail, but it genuinely makes life easier: the next time OpenAI ships something, you will know immediately whether it is a new generation or an upgrade to a tier you already use.

Sol, Terra or Luna: the difference in plain language

• Sol is the most capable model OpenAI has built, with new top scores on tool use and coding. It also gets a new “max” setting for extra deep thinking, and an “ultra” mode where the model puts sub-agents to work to solve large tasks faster. Price in the API: $5 in / $30 out per million tokens.

• Terra matches the previous generation (GPT-5.5) on performance, at half the price. $2.50 in / $15 out.

• Luna is the fastest and cheapest. $1 in / $6 out.

For most of the tasks a business actually does with AI (drafts, summaries, research, reports), Terra or Luna is more than enough. We make the same trade-offs every day in our own operations: heavy analysis gets the top model, routine work gets the cheap one. Once the AI starts working agentically and running tasks on a schedule, the token price is what decides what the automation costs you per month. At that point, a halved price is bigger news than any benchmark.

Why did the launch take two weeks?

OpenAI showed GPT-5.6 as early as 26 June, but only to a small group of approved partners. The reason is that the model has become so capable in cybersecurity that OpenAI coordinated the rollout with US authorities before opening up broad access.

That is worth pausing on. Frontier models now have capabilities that are assessed on par with sensitive technology, and they ship with layered safeguards: built-in limits in the model itself, real-time checks while answers are being generated, and account-level review when usage looks suspicious. As an ordinary user you will notice little of this, apart from the fact that some requests in gray areas may take longer or be blocked.

What does this mean for your business?

Three things, in practical order:

1. Everything you do in ChatGPT gets better without you lifting a finger. GPT-5.6 is rolling out in ChatGPT, and it powers ChatGPT Work, which launched at the same time. If you already have good routines with AI, they get a free upgrade.

2. Automation gets cheaper. Terra delivers the previous generation's top tier at half the price. If you are building workflows on top of the API (or paying someone to do it), the math needs an update.

3. The pace is the whole point. GPT-5.5 arrived just a few months ago. Now it has been overtaken and halved in price. Waiting for the pace to settle down is not a strategy. Building routines that survive a change of model is.

That last point is the real lesson. The companies that get the most out of this are not the ones that manage to pick the right model today. They are the ones that have documented their workflows, gathered their context and can swap engines without tearing the house down. The models will keep changing names. Your customer data, your tone of voice and your routines are what last.

The bottom line

GPT-5.6 is a solid upgrade, but the most important news is the structure: three clear tiers, a lower price on the mid tier, and a launch pace that rewards companies set up for change. If you have not started yet, start now.

Want help setting your business up so you actually get value out of every new model generation, instead of starting over every time? That is literally our job. Get in touch for a no-obligation chat.

Frequently asked questions about GPT-5.6

When was GPT-5.6 launched?

OpenAI showed GPT-5.6 in a closed preview on 26 June 2026, and made the models publicly available on 9 July 2026. The rollout in ChatGPT, Codex and the API is happening gradually.

What is the difference between GPT-5.6 Sol, Terra and Luna?

Sol is the flagship with the highest capacity, Terra is the balanced mid tier, and Luna is the fastest and cheapest. The number 5.6 tells you the generation, while Sol, Terra and Luna are fixed tiers you choose between based on need and budget.

What does GPT-5.6 cost?

In the API, Sol costs $5 in and $30 out per million tokens, Terra $2.50 and $15, and Luna $1 and $6. In ChatGPT the model is included in your subscription as usual.

Is GPT-5.6 better than GPT-5.5?

Yes. Sol sets new top results on coding and tool tasks among other things, and Terra matches the performance of GPT-5.5 at half the price. For most people the improvement shows up mainly in the AI solving complex tasks with less guidance.

Why was GPT-5.6 available only to selected partners at first?

The model has strong cybersecurity capabilities, and OpenAI coordinated the launch with US authorities. That is why they started with a limited preview for approved partners before opening public access.

Which GPT-5.6 model should a small business choose?

Start with Terra or Luna. They cover most tasks, such as drafting text, summarizing, research and reporting, at a fraction of the price. Use Sol when the task calls for deep analysis or long, complex workflows.