Memory settings

What is Memory, and How Does It Work? How Do I Adjust Memory Length?

Memory refers to the content the AI model retains during a conversation. We recommend using the default memory length for optimal performance and cost-efficiency.

In Advanced Memory Mode, memory length determines how many dialogue rounds the AI will remember (one round = your message + AI's reply).

Higher turn counts mean:

• More context is fed to the AI.

• Higher token usage (text volume), which increases Ruby costs.

Current AI technology still has limitations. Even with full memory input, the AI may occasionally "forget" details.

Click the + icon on the left side of the chat window.

Select the memory length you preferred, or use the "Advanced Memory Mode" if needed.

What’s the Purpose of Advanced Memory Mode?

We designed Advanced Memory Mode to let users have full controll over their chat experience. You can balance three key factors:

• Memory retention (how much context the AI retains),

• Chat costs (higher memory = more tokens/Ruby usage),

• Model performance (different AI models have varying memory capacities, all of which are inherently limited).

This flexibility lets you prioritize what matters most: richer context, cost efficiency, or leveraging a model’s specific strengths.

3. Pro Tips to Save Rubies

Method 1: Reset Chats (Keep Context, Lower Costs)

1. Make a Summary: Paste this exact command into your chat: (Execute command: Summarize all key events and memories <plot> to date)

→ The AI will generate a compact recap inside <plot> tags.

1. Migrate to a New Chat:

   • Click Edit on the summary, copy the text.

   • Start a new chat, paste it under the greeting like this:

     <plot>  
     【Previously】  
     [Paste summary here]  
     </plot>  

   then you can continute to chat with the AI in a new chat.

→ Fresh chat = lower token count, same narrative flow!

Method 2: Pin Critical Plot Points

• You can add pivotal details to your Persona (e.g., "The protagonist is allergic to roses"). This acts as a "cheat sheet" for the AI, hence reducing reliance on memory.

---

FAQ

1. If I’ve only chatted for 2 rounds but set memory to 100 rounds, will I be charged extra?

A: No. Costs are calculated based on the actual word count processed by the AI, not your memory length setting.

---

2. What happens if my chat exceeds 200 rounds?

A: Our context search model ensures relevance, even with limited memory settings. For example:

• If you chat 200+ rounds but only set memory to 30, the AI will prioritize retrieving critical early context (e.g., family details from the first few rounds) over less relevant later exchanges.

• The AI still recalls key details without needing ultra-high memory limits.

---

3. Why did my Ruby usage suddenly spike?

A: Increasing memory length when in a long conversation will dramatically increase the input tokens and the price. Example:

• Starting at 8 rounds → raising to 20 rounds forces the AI to process 20 rounds’ worth of text instead of 8.

• More tokens = higher Ruby costs. Adjust the slider cautiously during long conversations!