Longer Memory

Does your character keep forgetting what you just told them? Their favorite food, the pet name only the two of you use, the promise you made yesterday — Longer Memory is the feature that makes your character actually remember all of it, even as the conversation stretches on. It's not complicated. Stick with this guide and you'll have it all figured out in five minutes.

Think of an AI character's memory like the space on a desk. When the desk is small (= the memory capacity is small), you have to shove older papers off the edge to make room for new ones. So as a conversation grows longer, the character slowly starts forgetting the things you said at the beginning.

Longer Memory widens that desk. It increases how much of the conversation a character can "have in view" at one time (this is called context), so they can remember things from much further back.

Why this matters: The more context a character can hold onto, the smarter, more consistent, and more alive the conversation feels. As a rule of thumb, the more they remember, the better the experience.

Each plan comes with a default "desk size." The K after each number refers to the count of tokens (a token is roughly a chunk of a word or character). The bigger the number, the more the character can remember.

These are just the defaults. If you want your character to remember even further back, you can crank it up yourself with the Memory Dial below.

Heads up! Every plan comes with a base capacity, and anything beyond that draws on SP / your weekly allowance. We'll walk through the details below.

If the base capacity isn't enough, you can turn the Memory Dial to push your memory capacity all the way up to 1M (one million tokens). Like turning up the volume on a radio, you use a slider to decide just how deep you want the memory to go.

1. Open the Longer Memory tab in the sidebar of your chat screen.

2. Switch the feature On.

3. Slide the Memory Dial up as far as you'd like.

4. Before you send, check the estimate the app shows you, then send.

Good to know: The deeper you go with memory (= the higher you set the dial), the faster your weekly allowance and SP get used up. It's a bit like making your character read a longer "review note" with every message, and that extra effort comes at a cost. That's exactly why we always show the estimated cost on screen right before you send — no surprises.

The core of it first: chatting with the Default model is unlimited on every plan. Extra costs only kick in "when you go deeper with memory than usual" and "when you use premium settings."

• SP is deducted only once per message (your send + the AI's reply, bundled together).

• The heavier the settings (deeper memory, longer replies, bigger characters), the faster your weekly allowance / SP runs down.

• That's why we show you the estimated cost before you send. [Related: Story Points & Costs]

You never get locked out: Even if your weekly premium allowance or your SP runs dry, the Default model is unlimited for everyone, so the conversation never stops. Just lighten the memory a touch and keep going.

If jotting down every detail of your conversation sounds like a chore, try turning on Auto Mode. Your character will summarize the important bits in real time as you chat and tidy them into your memory notes automatically.

The maximum length a summary note can hold — and the cost — varies by plan.

Tip: On Free and Silver, Auto Summary uses SP, but on Gold and Platinum a larger summary capacity is built right into the plan. If you lean on summaries often, an upper-tier plan goes a lot further.

You know how there's always that one piece of info your character must never forget? Things like the dynamic between the two of you, the character's name, or a promise that can never be broken. Pinned Memory is the spot where you tack those down with a pushpin. No matter how long the conversation runs, this part never gets bumped off the desk.

Pinned Memory is available on the Gold and Platinum plans.

Why this matters: In deep roleplay, nothing is more frustrating than a character losing track of a core detail after a long conversation. Pin just the essentials to Pinned Memory, and that part stays alive the whole way through.

Don't panic. It's almost always one of these.

• "I increased the memory, but the cost looks the same?" → Make sure you checked the estimated cost before sending after raising the Memory Dial. The higher the dial, the faster the deduction.

• "I'm out of SP." → No worries. The Default model is unlimited, so the conversation keeps going. Lighten the memory a bit, or top up your SP to go deep again. [Related: Story Points & Costs]

• "I can't find Pinned Memory." → Pinned Memory is Gold/Platinum only. Check your current plan.

• "My Auto summaries keep getting cut off." → You've hit your plan's Auto Summary max character limit. Check your plan's cap in the table above, and consider an upper-tier plan if you need more room.

• Go long only where it counts. You don't need every conversation to remember all the way up to 1M. Usually the default, or one step up, is plenty — and it saves on cost.

• Put the essentials in Pinned Memory. If you're on Gold or Platinum, pin down the settings that can never be forgotten.

• Leave summarizing to Auto Mode. If writing things out yourself is a hassle, Auto Summary will organize the key points for you.

• Check the estimated cost before sending. Glance at the cost each time you raise the dial, and you'll never get caught off guard by a sudden SP drop.

• Longer Memory widens your character's "desk" so they remember further back.

• Base capacity is Free 8K / Silver 16K / Gold 32K / Platinum 64K, and you can raise it up to 1M with the Memory Dial.

• The deeper you remember, the faster your weekly allowance / SP runs down — but we always show the estimated cost before you send.

• Auto Summary (Gold 5K · Platinum 10K included) and Pinned Memory (Gold 1K · Platinum 5K) let you lock down your key memories nice and tight.

• And above all, the Default model is unlimited for everyone, so the conversation never stops.

Is something missing or wrong in this doc? Let us know anytime — just reach out to support and a real person will help you out.