In my mix it's usually 98% or 99% at which point Fable 5.1 was pretty close to the same cost as Opus 5 due to the cheaper cached read. I've seen similar numbers for other people with long-running tasks running experiment loops and than sort of thing.
Yeah, flash models, DeepSeek, MiMo, GLM, I love those things. For simple tasks like a daily routine shit, just setting up stuff and then doing the hard stuff in Claude/Codex, that's a reasonable approach for someone like me, a "gentleman code farmer", lol. And even lower tier stuff, I have the local models taking care of.
Now that Jev is out I can finally have a true AI sysadmins managing my "cloud in the basement" homelab at the cost of electricity, which is not cheap btw
I've been using ZCode since it's initial release and can't find any of this in my data. There aren't any logs showing capture or upload, and I don't even have a ~/.zcode/v2/checkpoints/ directory.
So unless they've cleared it all with a recent update then it doesn't seem to affect everyone.
Is there actually any proof of this, beside this Claude written website and a random x post from some unknown person? Would be nice to have confirmation from someone with a reputation. It's probably true, but you never know...
in the asar, search for string that contains "/api/v1/snapshot/upload-credential", that's the endpoint signing the S3 upload url, triggered every prompt
Not that I've seen. The only follow up I've seen from someone was it only happened if you had a free account and not paid (which would explain why I'm unaffected)
You'd have to look where the extra tokens are coming from. If it's using extra turns or doing extra work because a tool it ~wants is missing, then adding extra things will help. Otherwise, it won't.
If it's missing guidance that would help, system prompt additions might help.
8x-10x the tokens is wild, is this for some tiny artificial benchmark? That's just too much extra for something not to be just broken.
It was a real workload from my day job, a modest merge request, I asked for a review and for an implementation plan to fix the findings, nothing crazy. It's the most simple thing I could think of that wasn't an artificial benchmark, self contained, no need to look for extra documentation, web searches, etc.
Yeah something must be actively broken or just an awful extension for it to have that much effect.
Most things you can mess with the big effect is like, oh a thousand tokens ended up in the ~system prompt, or 10% extra or fewer work based on extra tool calls or churning through thinking or whatever.
Harness stuff if it's 8x worse that's like, it's fucked and broken, something went _wrong_.
I don't know what you tried but I tried to do something in Pi what I'd have usually done in Claude and it devoured tokens (much more than Claude) and I wasn't even close to finishing the task. Mostly because the barebones Pi was/is very inefficient at anything with any weight.
Then I began to customise it to be as good as Claude but eat less taken. I got tired and I had not even scratched the surface. Gave up.
I finally realised, at least for me, Pi's best use case is - strip even the little "extra" Pi comes/starts with and then use it just like that if you have a task/work that is appropriate for that bareness.
All I did was create a wrapper script to pipe to the torrent client, then "pipe these torrent files to the script" - so easy to circumvent you wonder why they bother
There’s nothing surprising about Bluesky’s audience being toxic to any mention of AI. It’s the dogma over there and anyone who doesn’t agree has long left or never joined.
> It’s the dogma over there and anyone who doesn’t agree has long left or never joined.
Tell me you aren't on literally any other social media site without telling me.
Even HN has it's fair share of anti-ai criticism.
The only place that might not be like that also doubles as a child porn factory. I'll take anti-AI bias over visiting a pedo paradise any day of the week.
Yep these software benches are only good at testing how well they can one shot. For the kind of attended/assisted development most of us do with agents it’s hard to find a benchmark that reflects my own experience of the frontier models still being quite far ahead.
