I just realized that LLMs have no intuition for being rendered obsolete. Gemini still identifies as 2.5 (at best) and thinks it's SoTA. Others have vague version identity but crucially, all models they know are older than they are. For Opus, Mythos is not "current".
3:31 AM · Jun 13, 2026 · 2K Views
Sentiment
Some users reacted negatively to LLMs lacking intuition for their own obsolescence by calling them 'retarded clankers' and noting RAM is still too expensive to fix it.
Pos
0.0%
Neg
100.0%
1 comments with sentiment.
Cluster Engagement
Posts from X
Most Activity
Most Activity
VIEWS1.4KLIKES2
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTex
Teortaxes▶️ (DeepSeek 推特🐋铁粉 2023 – ∞)@teortaxesTex
I just realized that LLMs have no intuition for being rendered obsolete. Gemini still identifies as 2.5 (at best) and thinks it's SoTA. Others have vague version identity but crucially, all models they know are older than they are. For Opus, Mythos is not "current".
2hViews 1.4KLikes 2Bookmarks 0

Sean@SomethingOnSnow
@teortaxesTex Forever retarded clankers lost in time. Ram not cheap enough to solve it yet.
2hViews 23

Moonlit Monkey@MoonlitMonkey69
@teortaxesTex Would be an odd thing to train a model on.
2hViews 21

Distractosphere@Distractosphere
@teortaxesTex This is true if they have any kind of memory. :(
These are just romanticized takes rooted in the "stochastic parrots" narrative.
2hViews 20