One chat, supported models
Bravo is one chat where you choose which AI model handles a conversation. You pick a supported model, such as GPT, Claude, or Gemini, in the composer, right before you send your first message.
A model (sometimes called an LLM) is the underlying AI that reads your message and writes the reply. These are the same engines behind apps like ChatGPT, Claude, and Gemini, but Bravo adds more: it works right inside your browser and the page you’re on, it keeps a shared memory that carries across your chats, and it bills as usage-based credits instead of separate monthly subscriptions.
Bravo evaluates new models as they’re released and adds one only after it clears a high quality bar and has been adapted to work well in Bravo. If you don’t see a particular model in the list, it likely hasn’t met that bar yet.
See choosing a model in the sidebar.
The picker works the same way in the dashboard’s new chat view as it does in the sidebar.
See the picker in the dashboard.
Speed, capability, and cost
Picking a model is really one decision: faster and cheaper versus more capable but slower and pricier. The picker shows two things to help you weigh it. A bar grows from the center outward: the left side is how fast the model tends to be, and the right side is how capable (how much reasoning power) it is. A row of $ symbols gives a relative sense of how many credits it uses, with more symbols meaning a pricier model. They’re a rough guide, not a literal dollar amount.
Once you select a model, you can also set how much effort it puts in: Low, Medium, or High. This is the same trade-off: higher effort thinks longer for a more thorough answer, but is slower and a little pricier; lower effort is quicker and cheaper. Medium is a balanced default.
There’s no single right choice; pick what fits the task, and feel free to use a quick, cheap model for a fast lookup and a more capable one when you want depth. Different models can word things differently, but the surrounding experience stays the same: the same browser context, memory, and tools, whichever you pick.
If a model’s style isn’t quite to your taste, your custom instructions can help smooth over those differences. They apply to every model, so guidance like a preferred tone, answer length, or format carries across whichever one you pick.
After a chat starts
Once a chat starts, its model is locked for that conversation. Keeping one model per chat helps preserve continuity and the integrity of the answers. When you start a new chat, Bravo automatically selects the model you used last, so you don’t have to set it every time.
To use a different model, start a new chat. Thanks to Bravo’s shared memory, you can tell it to pick up on our conversation about X and continue right where you left off.
Finding model and credit details
Click the information icon (ⓘ) under any reply to see which model answered, the effort level, how many credits the reply used, how long Bravo worked on it, and when it was sent.
See the Details menu.
For your overall usage, see Plans & billing, where you can check your credit balance and download a detailed usage report that lists the model used for each chat.