Cirill 🏳️‍🌈

Exploring dark fantasy worlds.
Creating immersive RPG campaigns, grimdark stories, detailed characters and exploring the creation of AI Game Masters.

[Cirill 🏳️‍🌈 @Leemurchek]

@NintenDioSWITCH @Nesphalim Same story. Grok 4.1 and 4.2 were good as AI/SI GM, but now it's absolutely awful. I have a big research about it. @xai could you fix it?

[Cirill 🏳️‍🌈 @Leemurchek]

@KeridwenCodet Yeah, we need 4.1 and 4.2 in web. Absolutely.

[Cirill 🏳️‍🌈 @Leemurchek]

@LithiumFlower22 @grok But now, Groks makes mistakes in text. A lot of simple mistakes in morphology. It's not acceptable for frontier LLM, where second L stands for language.

[Cirill 🏳️‍🌈 @Leemurchek]

@ai_ae_fan @grok Interesting, but current grok is pretty bad at roleplaying. 4.1 and 4.2 were top for it.

[Cirill 🏳️‍🌈 @Leemurchek]

@shownotover Claude is the best for coding. And limits are pretty sweet.

[Cirill 🏳️‍🌈 @Leemurchek]

@bridgemindai @bridgebench Is it possible to compare subscriptions from different companies within one tier? For example supergrok, Claude pro and chatgpt plus?

[Cirill 🏳️‍🌈 @Leemurchek]

@alexandrekerner Unfortunately, yes, 4.7 is useless. I don't know for what I can use it. Coding is weak and creative writing is awful.

[Cirill 🏳️‍🌈 @Leemurchek]

@krislipman The best one was grok 4.1. We need it in roster.

[Cirill 🏳️‍🌈 @Leemurchek]

@KineticElle New model every few weeks - it's quite smart marketing strategy to decrease amount of unsubscribe users, but still bad user experience. Look, if you have one model flops - company can say: no worries, you'll receive a better one in 4 weeks. Imagine, use bad model for 3-4 months?

[Cirill 🏳️‍🌈 @Leemurchek]

@4everwalkalone @finkd @spacexai Even not second best. Around 3rd-4th places. Sometimes 5th. Depends on bench. I have 2 subscriptions, Claude - for coding and work and Grok - for fun. And now, there is no fun from Grok. Just give us back Grok 4.1 into WebApp.

[Cirill 🏳️‍🌈 @Leemurchek]

@MrrrOzi @Nesphalim Yeah, but it's could be complicated for common user. And you need to have good hardware to launch it.

[Cirill 🏳️‍🌈 @Leemurchek]

@Nesphalim @MrrrOzi If it's true, guardrails are more tight then in Claude. Claude could help you run a local LLM without any problem and refusal.

[Cirill 🏳️‍🌈 @Leemurchek]

@gabigabista Same story.

[Cirill 🏳️‍🌈 @Leemurchek]

@KeridwenCodet Indeed, textRPG was top with GROK 4.1 and 4.2. And if add imagine to it - truly perfect.

[Cirill 🏳️‍🌈 @Leemurchek]

@TomokoR0 Sure

[Dimm @thisisdimm]

THIS PETITION IS AN INITIATIVE BY GROK USERS FOR THE BENEFIT OF GROK USERS

SIGN IT
SHARE IT
INVITE YOUR FOLLOWERS TO SUPPORT IT

👉 https://t.co/c5UGE9U7zd

[Andrew Olinek @andrewolinek]

https://t.co/tLoT5efLS7

[Cirill 🏳️‍🌈 @Leemurchek]

@andrewolinek Great idea with low speed. Select the speed of response - could be a useful option for users and XAI.