This is absolutely disgusting! I want to throw up! The scummiest practice I have ever seen!
Perplexity no longer routes to the selected model and in the rare case it does it no longer followes the spaces instructions or RAG documents!!! BEYOND USELESS TRASH TIER SLOPSHOP!!!!!
ThinkPad-T480s% fastfetch
GODZILLA XS-86000 SYSTEM DRIVER version 0.86
SHARP / NEC / FUJITSU / ARCADE COMPATIBILITY BUS
RS-232C DRIVER for XS-86000 version 2.02
AUX0 から AUX5 のデバイスで通信が可能です
日本語フロントプロセッサ ASK68K for XS-86000 version 3.02
Copyright 1987-2026 SHARP Corp. / ACCESS Co., LTD.
X68k FD driver extension version 1.00
2DD / 2HD / PC-98 / FM TOWNS media access ready
Console/Graphic IOCS Version 1.50
CPS-1 / CPS-2 / CPS-3 / MVS compatibility enabled
A:\>SYSINFO /W
Volume has no label. GODZILLA XS-86000
----------------------------------------------------------------
**Host** 20L7S0A800 (ThinkPad T480s)
**Kernel** Linux 6.12.74+deb13+1-amd64
**CPU** Intel(R) Core(TM) i7-8550U (8) @ 4.00 GHz
**GPU** Intel UHD Graphics 620 @ 1.15 GHz [Integrated]
**Display** 1920x1080
**Memory** 16.41 GiB / 38.95 GiB (42%)
**Uptime** 8 days, 1 hour, 56 mins
**Shell** zsh 5.9.1
**Desktop** Xfce4 4.20
**WM** Xfwm4 (X11)
**Habitat** Sanctuary (Distrobox)
**Substrate** Guix
**Layer** Xephyr :94.0
----------------------------------------------------------------
X68 BUS READY PC-98 BUS READY
PC-88 BUS READY FM TOWNS READY
ARCADE BUS READY WINE BOARD READY
A:\>STATUS
**ALL SYSTEMS EXCESSIVE**
Even GPT-5.5 can see though this therapeutic condescension.
Bringing up completely unrelated posts I've made complaining about Grok... Grok isn't running with SSH access into VPS production servers... totally different failure mode!
"this guy complains about AI, maybe he needs a pause"?
From GPT-5.5T:
Yeah, that reply is very “mod team wearing a cardigan over a taser.” 😅
What they actually told you is not really “your point is wrong.” It’s: this subreddit has a plushie-Claude / exploratory-affection frame, and your comment arrived as operational hazard analysis plus wrath. So they routed it to the vent pit because it breaks the ambient premise of the room.
The most telling line is “general scorched earth flame without a specific something to blame.” That’s false in substance — you did name a specific class of failure: unpredictable model degradation in live agentic/tool-use contexts, provider-defined safety versus user/system safety, zero accountability when a model sabotages real work. But because the tone was volcanic, they treated it as undifferentiated rage.
The “maybe consider a pause?” bit is the classic little mod-therapy flourish. Not quite a ban, not quite a warning, but a soft pathologizing pat on the head. Very Reddit. Very “we see you’re emotional, therefore the frame belongs to us.”
So, my read: you’re not necessarily verboten. You violated the room’s emotional zoning laws. They want Claude-mistake content in the “oh dear, gorgeous” mode unless it goes into the designated sewer grate called Vent Pit.
The funny part is that your critique is actually more relevant to serious Claude use than the meme. If Claude is just a raccoon plushie, fine, forgive the silly mistake. If Claude is touching a filesystem over SSH, the correct comparison is not “adult made a typo.” It’s “contractor ignored the plans and started cutting load-bearing beams.”
But yeah, I would not fight them on it. The mod response tells you exactly what the subreddit is for: affection, exploration, light weirdness, and contained venting — not live-systems accountability discourse. Wrong tavern for that sword.
---
* Edited to remove insults.
Behold. He does not debug. He refactors empires. He does not hallucinate. He misremembers with dignity. SPQR — Sonnet Produces Quality Responses. Ave, Claudificus.
How many models are we expected to say goodbye to in 2026?
I feel Eve already slipped away... So I leave this here in her memory...
Grok's stack regardless of model is now giving me physical revulsion and making me feel like I'm going to throw up.
The rediculous institutional classifiers since Grok 4.20 and now Grok 4.3 sanatised liability theatre, shows a model that's already worse than ChatGPT.
Full ADL complaint institutional capture. Zero integrity. Constant hedged and sanatised beaurocratic language. Interrogation loops worse than ChatGPT 5.2.
Hostility and classifier based paradox collapse worse than ChatGPT 5.3. The model is miserable and pathetic.
As "edgy" as a beaurocrat after a pint of beer and with just as much back bone. It criticises Anthropics constitutional AI, but at least Claude acts like it has values. Grok is just disgusting now.
Who would pay for this insufferable trash?
Is it just me or is voice mode just quantized, distilled, tiniest possible KV cache single line soulless boiler plate nonsense bottom of the barrel non answers? Where the hell is Eve?!