15 comments

[ 1052 ms ] story [ 4164 ms ] thread
I guess it shouldn't be surprising that Anthropic uses their own (really bad at writing) models to write their dev blog, but it sure is disappointing to read slop from a company that can easily afford human writers.
Wow this text is horrible to read and has all the AI smells. I'm quite surprised because Anthropic as a company has enough self awareness and sophistication to avoid this, it affects their brand after all.

Here are the AI smells

> You don't set out to buy millions of tokens. You set out to build a feature, finish a migration, or run a task. The token count is whatever the model needed to get there.

You can see how forcefully punchy and grandeur it reads. Pangram also flags it as 100% AI written.2

Forceful punchiness and grandeur is a great way to describe it.

I’m always surprised how hard it is to describe the obvious AI tells to non-tech people. My prior best attempt was “breathlessness”

i wonder if the people at anthropic are just so brainwashed by constant claude-ing that they’ve just started to talk like it as well
All AI character assassinations here without discussing the actual content: Opus 5.5 costs. I understand it is less fun to read GenAI messaging, but is there nothing of interest in the post?
Their provided calculator is of some use, but otherwise, the article has very little meat on it.

https://claude.dev/blog/what-a-task-costs-on-opus-5-5/#try-y...

What's discussed is the same that I've already heard from a dozen or so YT channels:

- Maximise cache reads to the fullest extent possible

- Keep to medium/high for almost all tasks. xhigh & max waste tokens.

- Regular compactions before reaching 200k context

Isn't there any way to use free claude Opus 5.5?
No but you get it on the cheapest subscription.
Illuminating article on pricing. I don’t understand how Anthropic is economically viable with unpredictable and potentially unbounded costs for business users. Businesses are going to be highly motivated to seek a more rational pricing. I certainly am. We are running Quen models for this reason.
Wait till you learn about how human employees have unpredictable and unbounded costs for tasks.
For budgeting purposes humans are predictable. They have fixed salaries that are easy to compute in advance and cost optimizations that are similarly not hard to understand.

Anthropic is SaaS on steroids at least as far as costs are concerned. It’s not entirely obvious how they defend their margins in a competitive market with customers who will be looking hard at bounding costs. If nothing else there’s a market opportunity for cost management products for AI.

Are we going to get ASI before we get a model that can write human sounding text? There's a kind of irony that these things were trained on the written word, yet are so terrible at it themselves.
OpenAI and Google's models are much better, even than Opus 5.5.