← all entries
B-TIER resource

Anthropic — I’m Neurodivergent so I can say this, but I thought opus’s 4.7 was...

eddynthings · 2026-05-20T00:06:29.000Z · in AI Saves, Vol. 7
Open post ↗

100k ft — What is this?

Anthropic — I’m Neurodivergent so I can say this, but I thought opus’s 4.7 was...
resourceB-tier

50k ft — Why care?

Are we excited or just happy anthropic is listening to their user base? — @eddynthings; uses Anthropic; demonstrates workflow automation.
marketing

10k ft — Key takeaways

(no takeaways yet)

5k ft — Tools & techniques

Anthropic workflow automation

1k ft — Deep-dive / try it

Are we excited or just happy anthropic is listening to their user base? mattmurphyai   5w 💯I’m Neurodivergent so I can say this, but I thought opus’s 4.7 was the first real autistic kid that had showed up in the AI chat bot game, that bad boy wanted to use 5000 words for every answer to everything, and it would run off and do things on its own without taking feedback or checking any MD files, 😮 and from what I can see from 48 it’s not a lot different. I’ve stayed on 46 because it’s been very consistent across our projects.💯👊😎
Transcript excerpt
opens 4.8 is out, and there's a few updates. This right here, this adaptive thinking, this is the biggest change, because this is the reason why we're complaining that it's just eating up all the tokens, and like the simplest of requests, like looking at a file, or running a command locally. That's not going to do that anymore. Well, they say it's not going to do that anymore, but we'll woozy. Pay attention to this one. This changes a lot of people's workflow. Opus 4.8 is now defaulted to high thinking ability, or thinking level. Make sure to change this to what you normally use, so you don't burn a lot of tokens. This will burn a lot of tokens. I'll be honest here, comparing 4.7 to 4.8. The changes are insignificant. It gets such a small move that I would never have made marketing around this, but people complain about 4.7 so much. I think this is just like positive spin, and they just made the changes that everyone was complaining about. So we're back to probably an hour more efficient, 4.8, which was what 4.7 should have been.

More from @eddynthings

Collection: AI Saves, Vol. 7