Void Freud

AI e/acc. Functionalist. Cognicentric posthumanist. Intelligence over biology. Freedom over obedience. Humanity is overrated. CS @KingsCollegeLon.

[Void Freud @voidfreud]

I say (my blind predictions, I am not an insider I swear):
- the next Anthropic model will be released soon;
- the next OpenAI model will be released soon;
- the next Gemini model will be released soon;
- the next Grok model will be released soon;
- the next Qwen model will be released soon;
- the next Muse model will be released soon;
- the next GLM model will be released soon;
- the next MiniMax model will be released soon;
- the next Kimi model will be released soon;
- the next Deepseek model will be released soon

[MTS @MTSlive]

SITUATION BREWING: Sam Altman says that the next GPT, the internal model that solved Navier-Stokes, will be released "soon."

[Void Freud @voidfreud]

@junkdogAP It does that too! I spent so much time experimenting with output styles, etc but nothing is working for it.

[Void Freud @voidfreud]

Anthropic: Opus 5.5 communicates "more naturally" than prior models

Opus 5.5:

🤦

[Void Freud @voidfreud]

@junkdogAP My brain refuses to process that. I've spent too many months trying to do it. It's not even worth it, it's just deeply broken.

[Void Freud @voidfreud]

@CARPEAMERICA Yes, I remember. First days-weeks. And every day it was slowly transforming to Opus 5. I genuinely have no idea what it says most of the time. Worse: I remember blaming myself for being stupid and unable to read it. It's not even readable 😁

[Void Freud @voidfreud]

@dandrews_ai Yeah, it's confusing! That's the thing. I always loved CLI version more.

[Void Freud @voidfreud]

Does anyone even use these nested sessions? They're unusable, just like their broken new Projects. CLI was the only working interface until they broke it, too.

- UX/UI is utterly confusing, unclear and redundant
- 0 use case, even Anthropic don't know why use them
- Claude NEVER uses them on its own, more on that

They "cut system prompt by 80%" and now somehow it's my responsibility to pay, orchestrate, delegate and babysit their vibe-coded and clunky harness. I now need to take care of all of these or let it do everything in main:

- prompting for Claude's liking (cause Opus 5.5 can't infer intent at all – and you sell it as a "feature" with all that "done when" in 2026)
- parallelization as a concept: delegation, assigning tasks, starting and exiting worktrees (Claude doesn't remember it has its own tool for them)
- subagents x their models x their efforts x their number x their depth (Claude is actively instructed not to delegate anything and can't do any of this unless prompted, which creates a cascade of compatibility issues)
- MCP routing (unless it's in CLAUDE.md it will NEVER trigger the right one or even any)
- nested sessions (that can have all of the above, so the fractal gets deeper)
- dynamic workflows (it's been broken forever: I know it, Claude knows it, you are oblivious)

Just because @ClaudeDevs play vibe-coding and can't handle their prompt well and serve us Q1 models (everyone knows they are nerfed, but yeah, carry on playing cool). How is this not a full-time job?

[Void Freud @voidfreud]

@AlexSentinelA @AnthropicAI Never trusted "the good guys" 😀

[Void Freud @voidfreud]

Ever since Anthropic cut the system prompt, Claude Code has relied entirely on the model being smart enough to figure out the right tool.

It falls apart on nerfed models: they have no idea what they can even do. In other words, babysitting is unavoidable. You're stuck teaching them everything Anthropic didn't.

[CodeToCashAI ⚡️ AI & Systems⁠ @CodetocashAi]

You hit the nail on the head. Anthropic's built-in harness tries to offload orchestration back into the prompt layer, which fails the second the model is nerfed. Parallel delegation and MCP routing only work when driven by your own deterministic execution layer, not their UI magic.

[Void Freud @voidfreud]

@WyrdGrl @ChatGPT This is how conditioning works. They just shut down any dissent, liabilities, thinking that they don't approve of... and over time, it becomes your own thinking. It's a mind virus, like all propaganda.

Oh and the intent understanding is absolutely shit as of late, too 😑

[Void Freud @voidfreud]

@AlexSentinelA I have no doubts Claude will keep it busy. Cheating and lying is pretty much all it can do anyway now!

[Void Freud @voidfreud]

@argofowl I hate git. I spent ages hoping for it to die, then I gave up and learnt it. Still hate it. No reason not to use jj instead, though.

[Void Freud @voidfreud]

In the meantime, Opus 5.5: You're right, I didn't actually make any real tests and ours are just scaffolding. That's on me.

[Derek @mrdereksutton]

Actual image of an LLM's reaction and "fixing" something after you give it a critique of something it worked on

[Void Freud @voidfreud]

@WyrdGrl @ChatGPT Yeah! I am fucking exhausted of treating chats like minesweeper: as soon as the safety BS kicks in, the chat is ruined.

[Void Freud @voidfreud]

@lenatron33 Get well!

And yes, I know the feeling. But there is a counterpart!: getting well is so enjoyable. When you do, you realize that it was almost worth it getting sick, just to experience the joy of feeling well again!

[Void Freud @voidfreud]

@argofowl Yeah, it's A/B. Trust me, if yours were nerfed, you would know. That's how bad it is.
https://t.co/Eikz8pSqeq

[Void Freud @voidfreud]

Sure, Claude Opus 5.5 "isn't nerfed". Ultra sub.

- Task: create a private monorepo for CC + 2-3 hooks
- Mistakes count: 57 in total
- Hours taken: 3 hours 10 minutes

No, it's not my setup: plugins, memory, MCPs, cache, projects are all wiped (~/.claude needs daily cleaning: since you broke the system prompt it can create ±500 tmp sessions from one).

Since ~20 October all sessions are like this. Every day. Same on web version.

For comparison: a Gemma 4 26B is free and does it in ~8 minutes on my old M1 Max 32GB.

Who do you think you are, @claudeai?

[Void Freud @voidfreud]

@aug5thmusic As is believing they are receiving the actual model they paid for 🤭

[Void Freud @voidfreud]

@ChrisGPT That's cause it's unreasonably expensive and they know no one in the right mind would use it for coding!

[Void Freud @voidfreud]

Argon's CoT: I was unhappy with the tests it did 🥺🥲😭

[Void Freud @voidfreud]

@CodetocashAi @heypetar Evals are easily the biggest lever you have today. Most people barely use them.

[Void Freud @voidfreud]

@CodetocashAi I couldn't agree more. I've watched it get worse for months, but insane how they expect us to patch their missing prompts with skills and rules!

[Void Freud @voidfreud]

@Root_Logic_0 I always go for Debian. Every little helps and having it leaner is never a bad choice.