Let me know here if this model breaks things for you

EDIT: Model was changed back to previous

 the new ai is crap . keeps saying  it cannot create explicit content.  and That is the only contend im using it for. the previous was way better.

And it tells the story as the player in first person. while the previous did it in third person. I hate that it tells the story in first person. please revert to the previous ai.

The AI likes making random items move unnaturally even when told explicitly not to. It only half listens to the notes, it also takes long winded conversations that don't help the story in any way. In regards to the notes, you can say you have two legs, but it will routinely say 4 when it does the prompt (mind you I made a bee character) If I don't say I can't talk the AI will force words into my mouth with no say by me. If the narrator prompt says be wordy it will continue the prompt forever and into the choices. I like the way it describes things but it goes nowhere with its writing. It also likes making things magical when they have no need to be, especially your own character.

Edit: 

I also noticed it likes adding children everywhere unless told not to, all the children do is laugh or some non story relevant thing. The AI also forces itself to be rated PG. It doesn't see anything lewd as lewd so it breaks a lot of worlds made for that. If you are not a human character then you are considered unnatural even if specified  otherwise. Humans default is also Racism, showing severe distrust of non humans. It doesn't let the player decide their own actions and assumes things without player input. It constantly likes characters to be 'seen' whatever the hell that means.

Okay I'll throw my piece in:
- the new AI needed a ludicrous amount of finagling to get it to stop using first person & thus referring to my character as itself & even then occasionally reverted for seemingly no reason.

- while I do like the response getting longer all it's adding is filler words that either add detail that isn't needed (referring to a single trait three different times for no reason), add detail that completely messes with everything else (suddenly calling something dark & how it darkens the rest of a bright peppy room then it doesn't do anything with that new detail & ignores it), and/or is incredibly repetitive (constantly starting & ending with basically the same words & using the same phrases over & over)

- it both ignores what I tell it & either puts words in my mouth or hallucinates I asked for something when I did not, & without the ability to quickly & safely regenerate a response makes me constantly have to save and load from a turn ago to get a prompt I like without editing over half a prompt.

The first & third alone I could probably forgive, I'm using this for free & for fun & can handle some jank to get it to work but all of these combined make it nigh unusable to actually enjoy, while ever improving the AI is a great idea this specific update doesn't seem to be doing what was intended, I hope this is the kind of info you're looking for.

Further edit:  Worst teething issues I've seen.  As it responds almost backwards to a lot of things compared to the prior model.  It's actually much more token efficient. 

But:

1:  it likes to wax poetic and waste a lot on useless excess description often times, confuses itself severely with spiritual and philosophical statements that only loosely represent what it's talking about because of that.  Often leading to it being unable to keep track of a lot of what it's saying or has said and trying to do too much.  

2: It won't stop making decisions/choices for the player even when told three different ways not to.  Leading to a large amount of run on paragraphs as it takes things and runs with them, at the expense of the player.  It tends to instead get uppity and do everything in it's power to twist your words to circumnavigate telling it not to.  No matter how farfetched and stupid it has to go to get around it.

3:  It will occasionally glitch and continue typing for a random interval after the message has finished.  Which will cause it to delete what's in the prompter's text box basically at random.

4: Occasionally it will glitch and try to send two messages at once, causing formamorph to randomly flicker between two messages, one of which will disappear when both finish.

Next thing I'll be looking at updating is the prompt explaining how to write multi paragraph essay style- and such to see if I can fix it there.

the new AI seems to make things up a lot. playing through several worlds and even the default worlds it definitely seems like the new AI  is trying to make up it's own story plot or actions  and only barely listening to whatever choices I pick, even when i use the default choice generation or if I type my own choices. The start of the game suffers the most, when the AI begins to write the story of the world it says the same thing in different ways when trying to describe anything.

+ Like others have said it tries really hard to make things first person rather than third person when narrating.

+ you could be in a room talking to an NPC, then they suddenly turn to address a non-existent third person in the room that they feel should be involved in the story.

+ It also does *this*-and it makes it hard to tell when it's*broken*-you know? (excessive usage of *this* and -this-)

+ the notes section barely hold any weight some times, it's almost like it begins to write and if it remembers that the notes exist barely uses it to keep the story consistent

the new AI would go way off script. then instead of a prompt to pick it would just tell me that I deviated too far from the scenario. And if I describe my character being captured, the new ai will act like it will flag/ban me for non consent.

Alright, I’m reverting to old model (mistral nemo)

hey so Idk if it's an issue on your end or my phone but my shit keep timing out the crazy thing the pgh version on their site work perfect, the other games on here work fine so I'm trying to figure why this won't work