un_commoncents

politically homeless

[un_commoncents @UN_C0MM0NCENTS]

@steipete hermes desktop is as bad as openclaw mobile.

[un_commoncents @UN_C0MM0NCENTS]

@AlexFinn dog shit > sonnet 5.

[un_commoncents @UN_C0MM0NCENTS]

@Teknium most days are good with Hermes but today it's dog shit.

[un_commoncents @UN_C0MM0NCENTS]

@Teknium @telegram Afraid to even turn this on. Hermes sucks at using telegram. Enabled threading and it’s a disorganized mess of responses in the new chat. After a day I correcting it I’m giving up going back to openclaw.

[Bob Starnes @bobstarnes]

If your OpenAI OAuth GPT-5.5 Hermes agent suddenly stops using memory/tools correctly while the same model works through other providers, check the harness.

In my case it was routed through Codex app-server, which bypassed the agent’s normal memory/tool loop and used a different tool universe.

Fix: switch OAuth runtime back to the normal Responses/Hermes path (`openai_runtime: auto`), restart the gateway, then test memory + tool calls in a fresh session.

[un_commoncents @UN_C0MM0NCENTS]

@theo Anti abuse = users so mad at the slop it puts out they were yelling at it

[un_commoncents @UN_C0MM0NCENTS]

@jordymaui I never once had this happen.

[First Squawk @FirstSquawk]

ANTHROPIC SHIFTS TO USAGE-BASED BILLING, INCREASING COSTS FOR HEAVY USERS - TIF

[un_commoncents @UN_C0MM0NCENTS]

@MeidasTouch only the paid actors made it.

[un_commoncents @UN_C0MM0NCENTS]

@Sthiven_R I couldn’t take it. Bailed out

[un_commoncents @UN_C0MM0NCENTS]

@petergyang 1.1 Gigawat doesn’t cut it

[un_commoncents @UN_C0MM0NCENTS]

@Jason Pass.

[un_commoncents @UN_C0MM0NCENTS]

@RichardGrenell Your prez said it's cool and he can rape all he wants.

[un_commoncents @UN_C0MM0NCENTS]

@DavidOndrej1 What happens when they don’t have the compute later when you need it?

[un_commoncents @UN_C0MM0NCENTS]

@steipete @aiDotEngineer Made my own hat.

[andrew chen @andrewchen]

built my openclaw a new home

(mac mini + nvidia dgx spark
in a 8U mini rack)

[Alex Finn @AlexFinn]

Something colossal is going to happen in the next 6 months

Right now every AI company on planet Earth is building AI agents for enterprise

Perplexity doubled their revenue the last couple months with it

Soon every enterprise will adopt them

When that happens, executives will quickly realize it can replace almost every low and mid level employee in the company

Anyone who has ever used OpenClaw knows this to be true. They know it's ALREADY better than them at almost everything

They know it's the most important software ever released

I think this is when the job losses accelerate

Humans at desks will be replaced by Mac Minis and Mac Studios

It has NEVER been more critical you are up to date on the latest AI tools

This is the ONLY way you'll be able to still have value through this chaos. If you know how to use the best tools, you can't be replaced by them

If you are an entrepreneur or creator with a platform you have leverage. You don't need jobs. You create your own value

I'd master these tools today:

• OpenClaw (duh)
• ChatGPT 5.4 (best coding model post Opus lobotomy)
• CapCut (so you can quickly pump out content and videos. Personal videos are the last way to be authentic)
• Local models (so you can have agents working 24/7 for you)
• And if you're daring: live stream. You can't AI generate a live stream.

The future is entrepreneurship. When there are no jobs, we will all be independent value creators

Start preparing

[Marcin Krzyzanowski @krzyzanowskim]

"Anthropic, OpenAl and Google release their new models with high quality from day one then slowly nerf them until the next model, so when the next model hits, its perceived as a bigger jump than it actually is"

sounds right what's happening

[Om Patel @om_patel5]

SOMEONE ACTUALLY MEASURED HOW MUCH DUMBER CLAUDE GOT. THE ANSWER IS 67%.

the data shows Opus 4.6 is thinking 67% less than it used to.

anthropic said nothing until the numbers went public. then suddenly Boris Cherny (creator of Claude Code) shows up on the GitHub issue.

users are calling it "AI shrinkflation" (same price, less intelligence)

we already know from the leaked source code that they have an internal switch that keeps the models working to their full extent for anthropic employees.

in the last week Claude went from WOW to being a more restricted and expensive version of ChatGPT.

people are saying Anthropic is deliberately downgrading Opus to save compute for training Mythos, their next model.

[un_commoncents @UN_C0MM0NCENTS]

@DarioAmodei Let me try it out I’ll fix everything in the world.