upvote

points

by cbg021 hours ago |
comments
upvoteby jstummbillig21 hours ago|
[-]
You are paying per token, but what you care about is token efficiency. If token efficiency has improved by as much as they claim it did (i.e. you need less tokens to complete a task successfully) all seems well.
reply
upvoteby mangolie21 hours ago|
parent|
next
[-]
Not for coding because it actually needs to read and write large files
reply
upvoteby baalimago20 hours ago|
parent|
next
[-]
Well, sort of. Imagine the case where it first scans the repo, then "intelligently" creates architecture files describing the project. The level of intelligence will create a varying quality of summary, with varying need of deep-scans on subsequent sessions. Level of intelligence will also increase comprehension of these architecture files.

Same principle applies when designing plans for complex tasks, etc. Token amount to grasp a concept is what matters.

reply
upvoteby jstummbillig20 hours ago|
parent|
prev|
[-]
Tbf, I have not super kept track of what is actually happening inside the "thinking" portion of recent releases. But last time I checked there still was a lot of verbosity and mistakes, that beat the actual amount of required, usable code generation by a wide margin.
reply
upvoteby cbg021 hours ago|
parent|
prev|
[-]
If it uses half the tokens to complete a task, then doubling the cost is perfectly fine. But is that actually true?
reply
upvoteby 2001zhaozhao21 hours ago|
parent|
next
[-]
This happens with every new model release though. The model makes less mistakes and spends less time fixing them, resulting in a token usage reduction for the same difficulty of task. Almost any task other than straight boilerplate will benefit from this.

In the same vein, I would guess that Opus 4.7 is probably cheaper for most tasks than 4.6, even though the tokenizer uses more tokens for the same length of string.

reply
upvoteby jorl1721 hours ago|
parent|
next
[-]
Maybe you'll have better luck but our team just cannot use Opus 4.7.

Some say it goes off on endless tangents, others that it doesn't work enough. Personally, it acts, talks, and makes mistakes like GPT models, for a much more exorbitant price. Misses out on important edge cases, doesn't get off its ass to do more than the bare minimum I asked (I mention an error and it fixes that error and doesn't even think to see if it exists elsewhere and propose fixing it there).

I've slowly been moving to GPT5.4-xhigh with some skills to make it act a bit more like Opus 4.6, in case the latter gets discontinued in favour of Opus 4.7.

reply
upvoteby cbg021 hours ago|
parent|
prev|
next
[-]
Doesn't look like it's cheaper, better or uses fewer tokens: https://www.reddit.com/r/Anthropic/comments/1stf6fz/one_week...

YMMV, I know.

reply
upvoteby user342837 hours ago|
parent|
prev|
[-]
Based on my experience with Claude Code on the $20 plan I would not think so.

Opus 4.7 would blow through the session limits in 2-4 prompts. It was a noticeable further decrease in usage quota, which was already tight before.

Based on Anthropic‘s description 4.7 was trained to think longer.

With GPT 5.5 yesterday, I felt it completes task noticeably faster than 5.4. I kept the xhigh effort setting.

reply
upvoteby jstummbillig20 hours ago|
parent|
prev|
[-]
We'll find out!
reply