Which Gpt model and reasoning level did you use in Codex and Amp?

Generally I have noticed Gpt 5.2 codex is slower compared to Sonnet 4.5 in Claude Code.

Amp doesn't have a conventional model selector - you choose fast vs smart (I think that's what it is called).

In smart mode it explores with Gemini Flash and writes with Opus.

Opus is roughly the same speed as Codex, depending on thinking settings.

This works: $ZAI_ANTHROPIC_BASE_URL=xxx $ZAI_ANTHROPIC_AUTH_TOKEN=xxx

  alias "claude-zai"="ANTHROPIC_BASE_URL=$ZAI_ANTHROPIC_BASE_URL ANTHROPIC_AUTH_TOKEN=$ZAI_ANTHROPIC_AUTH_TOKEN claude"

Then you can run `claude`, hit your limit, exit the session and `claude-zai -c` to continue (with context reset, of course).

Someone gave me that command a while back.

thats pretty much what I do, I have a bash alias to launch either the normal claude code, or the glm one

You can still test it for accuracy?

ah, the thing every person does when searching for a calculator: verify that it can actually do the math, a thing computers were historically good at from about the 1960s until within the past few years.

do you understand how bad it is when you search for software and you cannot trust it to do what you ask of it? it's bad!

Imagine saying this for medicine.

> consider a scummy way to make money.

How should Chatgpt survive then?

Well, the obvious answer is none of these companies should.

ok but the real world does not work like that.

The one thing it will never recommend is no product at all.

You've ceded even the glimmers of discernment that remain in you and all I feel is pity. It is not a 'waste of time' to interrogate your own desires.

There's no such thing as a 'top 3' for all things under heaven. You cannot purchase yourself a solution to every 'use case'. Furthermore, even if there were such a ranking, the ad machine would not reveal it to you, as you are not the customer, you are the mark.

You don't even want to be bothered to hit the 'buy it now' button. This is the mental model of an immaculate rube. You deserve better.

But this loses value as soon as ads come into play because of incentives. It's hard to trust recommendations when ads are in the mix.

> You can't attract world-class talent in research with a culture built around paranoid secrecy.

Would giving more money/shares help?

> Challenge the law in court

Do the courts in Italy work or do they do what the govt wants them to do.

The government complains everyday about the judges and it's trying to make a referendum to make judges angry, so I wouldn't say courts do what govt says

Yes, the judiciary is an independent branch in Italy, alongside the executive and the legislative.

> but it obviously doesn't make sense if you think about it for more than five seconds.

Maduro would disagree.

Its faster and the coding models before 5.2 for me did not work at all. Claude models have been reliable since 3.5

I dont think WSJ makes money from subscription fees, ads keep the lights on for journalism.

It's not that hard to find the data. Subscription fees are the great majority of WSJ overall revenue. ~$500 million of $586 million total.

https://www.amediaoperator.com/news/wsj-subs-rise-as-pricing...