ZeroTwo offers six OpenAI models from the GPT-5.6 and GPT-6 families. Each name ends with a tier word: Sol, Terra, or Luna. Which of them you can pick depends on your plan, and what they cost depends on the plan too.
The six models
| Model | Available on Free | Credits in regular chat on paid plans |
|---|---|---|
| GPT 6 Luna | Yes | No credits on Pro, Pro 2×, Ultra, and Ultra 2×. Plus and Business use credits. |
| GPT 5.6 Luna | Yes | Uses credits on every paid plan. |
| GPT 6.1 Sol | No | Uses credits. |
| GPT 6 Sol | No | Uses credits. |
| GPT 5.6 Sol | No | Uses credits. |
| GPT 5.6 Terra | No | Uses credits. |
All six have a context window of about 1 million tokens. Plus, Pro, Pro 2×, Ultra, Ultra 2×, and Business can use any of them. A Free account that holds purchased credits can also use any of them and pays as it goes. Without purchased credits, a Free account can use only the two Luna models from these families.
ZeroTwo meters chat in credits on every plan, Free included, except for models that cost no credits on your plan. To see where the credits go, hover over the coin icon on a model's row in the picker. It shows the model's estimated cost in credits per 1,000 tokens. The icon is missing for models that cost no credits on your plan. For more on credits, see How do credits work in ZeroTwo?
Choose a model
- Select the model name in the composer to open the model picker.
- Type in the Search models... box to find a model by name or provider, for example "Luna".
- Select the model you want.
You can also choose the chat model in Settings > Personalization > Models, using Chat model. The picker also has an Auto row. Auto is not a provider model. It sends your message to GPT 6 Luna.
In the web and desktop apps, the Help menu has a Model Tiers entry. It groups models by credit cost and shows the limits for each plan.
Thinking effort
For GPT 5.6 Sol, GPT 5.6 Terra, GPT 5.6 Luna, GPT 6 Sol, and GPT 6 Luna, the composer has a thinking-level button. It shows the current level, such as Standard, and screen readers announce it as Thinking effort. What it offers depends on the mode:
- Chat: None, Standard, or Extended. None answers right away without thinking. Standard corresponds to Medium and Extended corresponds to Max.
- Work and Code: the model's own steps, Low, Medium, High, X-High, and Max. These modes always reason, so there is no None.
These models start at Medium, which Chat shows as Standard, until you choose another level. GPT 6.1 Sol always reasons and has no None setting, and the composer does not show a thinking control for it.