No daily cap. No token quota. No paywall mid-scene.
The worst moment in a hosted AI chat is the one where the conversation is working and a banner tells you to wait until tomorrow or upgrade. That limit isn't arbitrary, every message you send costs the host real GPU time. privstory doesn't have it, because the GPU time is yours.
Why hosted chats meter you
Running a language model costs money per message. A hosted service pays that cost for every user, so it has to bound usage somehow: a daily message count, a token budget, a credit balance, a slower model for free users, or a subscription tier that raises the ceiling.
Those mechanics are a direct consequence of the model living on their hardware. No amount of goodwill removes them, because the marginal cost of your conversation is real and someone has to absorb it.
What changes when the model is local
The chat model runs in a process on your machine, using your GPU or CPU and your electricity. The marginal cost of a message to anyone else is zero, so there is nothing to meter and no reason to interrupt you.
There is no counter to watch, no cooldown, no notion of a conversation being “too long” for your tier, and no difference in model quality between your first message of the day and your thousandth. Scenes can run as long as you want them to.
The one real limit: your own hardware
Being honest about the trade-off: local generation is bounded by the machine you run it on. On a recent Apple Silicon Mac or a discrete GPU, replies stream at a comfortable reading pace. On integrated graphics they are slower, and on CPU-only fallback image generation in particular is slow.
That's a speed limit, not a usage limit, and it doesn't reset at midnight or ask for money. If you want faster, the answer is a smaller model or better hardware, both of which are your call.
How the free version differs
The free version limits how many characters and scenes you can have, one of each, not how much you can say in them. Message volume is unlimited on both the free and licensed versions.
A one-time license lifts the character and scene limits. It doesn't add message capacity, because there was never a message limit to lift.
Common questions.
Are messages really unlimited on the free version?
Yes. The free version restricts you to one character and one scene, but within that scene you can send as many messages as you like. There is no daily cap, no token budget, and no throttling.
Does the conversation get worse or shorter as it grows?
Very long scenes eventually exceed what the model can hold in context at once, which is a property of the model in your module rather than a billing limit. Different modules ship different models, so the practical ceiling depends on which one you choose.
Is there any per-message cost at all?
None to us. The only cost is the electricity and the time your own machine spends generating the reply. Nothing is billed, counted, or reported.