A free guide by New Money School
New Feature
OpenAI just released GPT-5.6 as three separate models. Sol, Terra, and Luna each have a different job. Here is when to use each one, and the one that quietly won me over this week.
OpenAI just released GPT-5.6, and the smart part is how they packaged it. It is not one model that tries to do everything. It is three, and each one is built for a different kind of job.
Here is the one idea that makes all of it click. The number, 5.6, is the generation. It tells you how new the tech is. The names, Sol, Terra, and Luna, are the tiers. They tell you how much power you are paying for. So next time OpenAI ships 5.7, you will still have a Sol, a Terra, and a Luna. Same three jobs, newer brains. That is the whole system, and it rolled out broadly on July 9, 2026.
What All Three Share
Before the differences, know the common ground. All three read up to a 1 million token context window, write up to 128,000 output tokens in one go, and know the world up to a February 2026 knowledge cutoff. The only things that change from one to the next are speed, smarts, and price.
Source Details from the official announcement. I also tested it myself.
The prices tell most of the story, so start there. Here is what one million tokens costs on each model. Input is what you send in. Output is what it writes back.
$5 / $30
Sol. In / out per 1M tokens. The powerhouse.
$2.50 / $15
Terra. In / out per 1M tokens. The everyday one.
$1 / $6
Luna. In / out per 1M tokens. The fast one.
Same generation top to bottom. The gap is what you pay for more thinking. Now here is what each one is actually for.
SolThe Powerhouse
Sol is the one you call when the answer has to be right. It is the smartest of the three, and it takes its time to think. This is the model for the hardest problems: deep research, complex coding, strategy work, and security research. You do not reach for Sol to clean up an email. You reach for it when a wrong answer would cost you real time or real money. Best answer, not fastest.
Now the part that surprised me. In my own testing this week, Sol is roughly tied with Claude's best model on raw intelligence. The two trade blows on the hard stuff. But Sol gets there at about a third of the cost. If you have ever been deep in a project and hit a usage limit right at the good part, you know exactly why that matters. Cheaper thinking means more thinking before you run out of room. That is why Sol is the one I now reach for more than anything.
TerraThe Everyday Model
Terra is the balance, and it is the balance in the best way. Fast enough to run at real volume, sharp enough that you trust what comes back. This is the workhorse for everyday business tasks: customer support, internal tools, and reading and sorting documents at scale. When you need good answers and a lot of them, and both speed and smarts matter, Terra is the pick. For most people, most days, this is the default.
LunaSpeed And Cost
Luna is built for speed and price. It is the fastest and the cheapest of the three. Hand it the simple, high-volume stuff: summarizing, first drafts, and routine automation. It will not out-think Sol, and it is not trying to. Its whole job is to be quick and nearly free so you can run it all day without watching the bill climb. When the task is small and you need it done now, Luna is your model.
Cheat SheetWhich One, When
Here is the whole guide in three lines. Match the job to the model and you will almost never pick wrong.
- Deep research, complex coding, strategy, or security work. Use Sol. The best answer wins, and it is cheaper than you would guess.
- Everyday business work at volume, like support, internal tools, and document analysis. Use Terra. Smart and fast, your daily default.
- Quick, simple tasks, like summaries, drafts, and routine automation. Use Luna. Cheap and instant, run it all day.
The Real Skill
Think of AI models like streaming services. If you only have one, you miss a lot of good stuff. Having more than one, and knowing which to open, is the actual skill. GPT-5.6 just made that simple. Same generation, three tiers, and now you know which one to open for the job in front of you.