Yumina

Community · Feedback

Context Memory Session Memory & Story Summary (Beta)

Swuell ·

Is this a bug where there's no way to adjust it in a world--like sometimes the option to even adjust this extension is literally missing and also there's no way to adjust the AI model in certain worlds. I'm not sure what the issue is or maybe i'm missing some kind of controls but it would help alot of the DEV team made a note about how this adjustment works: since from the wording it seems like it should just appear in the chat/world but I'm assuming that's not how it's supposed to be? If it needs to be added that info should be added as a huge notice and caveat so that players and creators can get a better understanding of this extension.

Also there's two things: the FAQ states that you can freely change ai models inside the chat through the interaction inside the chatbox but i've found that to not be true for 100% of the worlds. Ones with heavy ui bundle assets and custom ui sometimes has no options of even changing the ai model. It's literally just not there. Is this an option that NEEDS to be implemented and it's not there by default code? If so this info should be told as well.

The second for the Session Memory extension can we please have it at least it default to free if you're not going to have the option of adjusting the extension outside of the chat but within chat only? Because I find it a bit egregious when it defaults straight to gemini 2.5 Flash lite when the website so far has promoted--and also been pretty transparent--about the upfront cost of the data model. It's just that some people may not check after installing the extension which means the extension would be running through gemini 2.5 and eating their allowance without their knowledge.

You could say it's up to the user to check but this isn't even noted within the extension. So someone very new to any type of ai chat/programming may very well not know since it's never even pointed out, let alone actually shows you where the session memory is--since the session memory is actually re-titled as "Context Memory"--so it can very well throw people off that are specifically searching for the aforementioned extension name.

But going back to the default, it means I have to manually hunt down the option for the extensions and manually change it to free each time I open a chat. Which is how it relates to the previous problem since that option doesn't always necessary appear in every single world instance. So it makes me not play it at all since I have no way of knowing if it's actually there just running without a ui telling me or if it's actually not even in the chat at all.

If a dev could please reply to this and try to fix this I would very much appreciate it. It would also be nice to know if it needs to be implemented or not. Also how to change the ai model if that option isn't even available in the said chat/world.

Thanks.

4 replies

  1. Yumina · Staff ·

    Hi thanks for this post. For your first question, for the worlds that uses our default template there is a model switcher option. And this is true for most of the cases. Yet, it is up to the author themselves to decide what UI they want - after all, it's their game. So, some authors might not include a model selector pill inside and there is nothing we can do about it. In terms of the memory extension, you are right and we will add a notification warning token usage when its first activated. However, it should also be understood that Gemini 2.5 flash lite is already one of the cheapest models and best of its price. If we switched to a worse default model, the quality of it will be very very poor. And finally, we have to emphasize that we always encourage people to bring their own API and understand that there is a cost of using AI - and it's not us that charges you.

  2. Swuell ·

    No I understand that completely and Gemini 2.5 flash is cheap; I'm just saying for others as a baseline you should just set it to the default free so that others won't get caught off guard. Because many will honestly. I wasn't blaming you about the cost, I know how the llm works. I was saying that to set it at gemini instead of the free is a bit egregious is all. It would just be better business wise to set it at the baseline of free and people can adjust it themselves. I would even propose to note it in the section--a detailed section--that the baseline is default as free but you can change it to whichever you want--and recommend Gemini 2.5 flash as the best/most cost effective alternative. There is quite a bit difference between the two since what I talked about actually just happened to me in an world. Since I didn't know it would count a generation on the first given message in a world--didn't realize the world had no first greeting and when i noticed it was too late--but due to that I was able to compare the results from gemini 2.5 flash lite vs. yumina on the context and it does do it a bit better for a very tiny amount of the cost.

    I would also note that you guys round up the cost. since the cost for me was 0.2 which would be 1199.98 but it ended up cutting into it 1199 instead.

    I just wanted to make sure that if it's not there then the extension, and default option isn't there so it won't count against you. Since this part about the creator needing to add it was and is…

  3. Yumina · Staff ·

    Thanks for the feedback. We will have a major update really soon and are currently working on updating the docs and FAQ page. And we are looking at the extension right now to see how it can be improved.

  4. Swuell ·

    Oh ok. I just thought you guys should know was all. Since I thought that's basically what the feedback part was for. You guys are doing really great though. So sorry if I seem like I'm nitpicking, but I'm really not. It's just the user friction that I'm addressing. Thanks for the response back as well.