How do I pick a model?
Every journey in Imagina is written by an AI model, the storyteller. Each model has its own writing style, speed and cost, and you choose which one writes for you.
Every journey in Imagina is written by an AI model, the storyteller. Each model has its own writing style, speed and cost, and you choose which one writes for you.
While Imagina is in beta, we offer a wide range of models on purpose. What works for one player may not work for another, and the same goes for different stories and worlds. Try a few and see what suits you.
Where you pick a model
- When starting a journey: the character picker has a model button.
- During play: open the journey tools and choose Model ().
- Your default: in Settings, Play tab, set a Default model. New journeys start with it, and you can still pick another model when starting any journey. Use Reset to go back to the platform default.
Models carry the real names their makers gave them. The lineup changes often, so this article describes how the picker works rather than listing specific models.
The picker
At the top is a search box: search by model name, maker, or writing style. Below it, tabs narrow the list:
- Favorites: the models you starred. This tab appears once you have starred at least one model.
- Recommended: good picks for most players and worlds.
- Standard: everything that is not Premium.
- Premium: the pricier models.
- Early access: only shown if you turned on Early access models in Settings, Play tab. These models are untested, can break stories or fail mid-turn, and can be removed at any time.
- All: the whole list.
Each row shows the maker's logo, the model name, a few badges and a cost meter. Long lists are grouped by maker, and deprecated models are folded away under Show N older models at the bottom (they open automatically when you search, or when your current model is one of them).
The picker opens on the tab that contains your current model, scrolled to it and marked with a check.
On a computer: hover a model, or move through the list with the arrow keys from the search box, and its details appear in the panel on the right. Click a row, press Enter, or use the Use button in the panel to switch.
On a phone: tap a row to pick it. Tap to open that model's details in a sheet from the bottom of the screen, with the same Use button.
Star a model () to add it to Favorites. Favorites are saved to your account, so they follow you across devices.
The details panel
For each model, the panel shows its maker, its badges, a short description, the price and any caveats. Descriptions cover:
- how the model writes,
- how it handles adult content where that matters (some stay tame and non-explicit, some write explicit scenes, some will not write explicit scenes at all),
- whether it thinks before writing (slower, but more consistent turns),
- known reliability problems.
Badges
- default: the platform's default model.
- a promotion, such as "40% off until Oct 8": a temporary lower price. The standard price is shown struck through next to the live one.
- recommended: a good pick for most players and worlds.
- premium: noticeably higher cost.
- SFW only: will not write explicit scenes.
- experiencing issues: currently having problems, and turns may fail.
- deprecated: an older model that may be removed.
- early access: see the Early access tab above.
A model that is temporarily unavailable is grayed out and cannot be picked. Its turns are redirected to another model, named in the note.
Cost
The cost meter on each row has four bars and shows a relative price band, from low cost to very high cost. It is a quick comparison, not a price. A fully free promotion shows Free for now instead.
The details panel shows exact prices in credits per 1 million tokens:
- Input: the story context sent to the model each turn.
- Cached input: context reused from the previous turn, billed at a lower rate.
- Output: what the model writes.
On a phone, the row also shows the input and output price.
What a turn costs depends on the model and on how long the story is. The context grows as the journey goes on, so later turns usually cost more than early ones.
Some models carry a pricing caveat in yellow: for example, the rate can rise once a turn's prompt gets very large, or the model can fall back to a more expensive provider. Check it before picking.
See also prompt caching and credits.
Which model should I pick?
Just want to play? Stay with the default model. It is picked for you, and you can change it any time.
Not sure where to start? Browse the Recommended tab and read the descriptions. They tell you how each model writes and what to expect from it.
Watching your credits? Compare the cost meters. Fewer bars means cheaper turns, which matters more as a journey grows.
Want the deepest storytelling? Look at the Premium tab, or at models that think before writing. Both cost more, and thinking models are slower, but they tend to keep long stories more consistent.
Found something you like? Star it () so it lands in Favorites, and set it as your default in Settings.
Switching models mid-journey
You can change models between turns without starting over. Your story, characters and world state carry over. This is handy for trying a different writing style partway through, or moving to a cheaper model when credits are running low.
Switching clears the prompt cache, so the first turn on the new model costs noticeably more. Later turns go back to normal. The picker shows this warning during a journey.
During a journey, Advanced model options at the bottom of the picker lets you:
- adjust the model's thinking level (on models that support it),
- turn thinking off for much faster turns (on some models),
- write a custom prompt for that model: your own instructions, added to the storyteller every time you play with that model, in any journey.
Custom prompts can also be managed in Settings, Play tab, under Custom model prompts. They are for experienced players: a badly worded prompt can hurt story quality.
What is the model cache?
The model cache (providers usually call this a prompt cache) is a short-lived copy of your story context. When it is warm, the next turn can reuse most of that context and usually costs less. When it is cold, the model reads everything...
Can I play in my language?
Yes. When you start a new journey, pick your output language from the language selector next to the model picker. Available languages: