Interesting perspective

ang benchmarks against state of the art?

It depends. You can expect a 5 to 15% performance hit depending on the task. In OpenBSD, security comes first and performance comes second.

> The Service is provided "as is." We are not liable for downtime, data loss, or errors in token counting or cost calculation.

lmao

source?

> When I come back to Slack, replies are often already sitting in drafts.

He must be a pleasure to work with

What makes you think so?

There's a certain writing style, very short paragraphs, fair amount of repetition that just feels like you've read this post before. And you have, just on different topics but it's always the same feel.

But also lots of negatives to start the sentence, usually with a reinforcement e.g.

your password didn't help. your 2fa didn't help. you were never asked to authenticate. you were asked to authorize. completely different mechanism

lmao at dang getting downvoted

The article has hallmarks of being formulated by an LLM. Why should I bother to read it if I ca not be sure which parts are based on the prompt, and which parts are hallucinated from the LLMs world knowledge? Dear author, care to simply share your prompt with us?

It wasn't "a prompt" but several prompts that transformed the raw experimental results to a blog.

> hallucinated from the LLMs world knowledge

This can't be true because I checked whether the content was consistent with the experimental outputs

The topic is interesting and you have my thanks for taking the time to look into it and prepare the post. Would you say it's fair to say that if you didn't use LLMs to prepare the post, we would have no blog post at all? In that case, I think I lean more towards being OK with this usage of LLMs, as I'd rather have this content available than not. However, I can only read that one repeated sentence about "booleans" (Ctrl-F "Boolean" and you'll know what I mean) this many times before I start questioning the validity of the entire document. It is not _good_ writing, to be frank.

lmao at opus 4.7 being a downgrade

They made it less sycophantic. Which is a good thing for mental health, but maybe a bad thing for popularity contests.

The question is ambiguous. It does not state that I want to wash the car there.

Why is the answer graded therefore on whether I took the car or not?

Maybe I wanted to meet a friend there?

GPT-5.4 might even price this in:

> "If somebody asks if they should walk or drive to a car wash 50m away, they probably don't need to actually wash the car there, otherwise the question doesn't make sense, they probably just need to get to that place efficiently. So walking it is."