Jeevo picks the right model for each message, so you never have to. But you can always see the whole list, pin any model yourself, and know precisely what each one charges. No subscription: you top up and pay for what you use.
It works, it's in daily use, and it still has rough edges. Models, prices and features change as we learn. Every bug report, every "that was confusing", every "I expected it to do X" makes it better — testing and feedback are genuinely appreciated.
15 messages a day on every model, chosen for you exactly as a paying user's are. Plus a taste of the rest: an image, a song and a document, and 2 days of the Today concierge reading your mail.
Top up by card, in your own currency. Your balance is spent message by message and nothing renews. The app shows a running total for every conversation, and prices every image, video and song before you confirm it.
Every model costs what its provider charges, plus one fixed margin that is the same on every model. We don't mark up the expensive ones more, and we don't make money by routing you to them.
Bring your own key for Anthropic, OpenAI, Google, ElevenLabs or Runway and you pay the provider directly — we charge only the 17.6 %.
Language models charge per token — roughly three quarters of a word. You pay for what the model reads (your message, the conversation so far, anything it looked up) and for what it writes back. A million tokens is about 750,000 words. Longer conversations cost more per message, because the model rereads them; re-reading is billed at the cheaper cached rate where the provider offers one.
Where a provider trains its models on what you send, the model is marked. Every model in the app carries the same labels, so you can see it before you pick it.
These are priced per item rather than per token, and Jeevo always shows you the price and asks before it makes anything that costs more than a message.
