I feel like this benchmark reiterates my disbelief that anyone uses the latest Anthropic models for any productive work. They seem to be the best at burning tokens and spawning unnecessary subagents even for well-defined and tightly scoped tasks.
Can we get a count of people that have had Claude read irrelevant documents or perform unnecessary web searches even when told not to from the beginning?
I'm starting to wonder if this increased token usage is inadvertently bleeding into how Anthropic actually trains their model, especially leading up to IPO. As older models are deprecated and users are forced onto newer models, if the default is less efficient and more token expensive that directly results in higher "profit" for Anthropic in terms of the consumption their users have to tolerate - lest they jump to a competitor.
I switched from Anthropic to OpenAI after spending ~$40K in equivalent token costs using Claude over 3 months.
I found Opus 4.7 to be slow and wasteful with token usage. It's shocking how inefficient it is with tasks like bash tool usage and web searching, delegating them to a dozen subagents only to get stuck and never return until you esc and intervene. That, in addition to all of the broken tooling Anthropic built in to limit token usage like the broken monitoring tool made managing Claude a chore. I was happy to pay $200/month for Opus 4.5 when they had more capacity, but 4.7 felt like a huge step back and no longer worth the price and inconvenience.
I remember an OpenAI employee comment on the GPT5.5 release post about how they specifically geared it towards long-horizon tasks and its been a breathe of fresh air in that regard. I have five two-week long sessions going right now and there's been no degradation in performance or efficiency. It's much better at carrying rules/learnings forward even in long-running sessions and grounding/refreshing itself in verified facts when it loses context.
Its funny because in two weeks I've gotten way more done with GPT5.5 with way fewer tokens and way less handholding. I think this goes to show how important tooling and the harness is and how a capable model like Opus 4.7 can be severely handicapped by bad product decisions.
The longest route on the NYC subway is precisely when you have an appointment and the train decides to stop because there's electrical issues, someone jumped on the track, "there's a train stopped ahead of us", the express decides to go local instead, someone is holding the door, your route involves the F/G or any line that serves less affluent neighborhoods...
Idk, my recent experience with Claude is that 4.7 barely knows how to use basic bash tools - how to properly check when programs have finished running, even basic stuff like how to run pytest suites and read the failed tests from the output without re-running the suite to specifically look for them. It's shockingly dumb for all of the tooling they've built into Claude Code (the useless Monitoring tool that blocks bash polling/sleeping that actually works, etc.).
I finally get fed up and started using GPT 5.5 the past 4 days and its a breath a fresh air despite feeling much more minimal. With Claude I had to write so many hooks to enforce behaviors it wouldn't remember and it lacked common sense on. GPT 5.5 does a much better job with things like knowing the AWS CDK CLI can hang on long CloudFormation deployments and it should actively check the deployment status using CloudFormation API rather than hanging for 30+ minutes - and it does this all without asking.
Maybe there's better tooling built into Codex too, but at least on the surface level it seems like how smart the model is makes a significant difference because Claude has more tools than I can count and still struggles to use "grep".
Edit: Like just now - I can't tell you how many times I day I see this sequence:
"Sorry, I'll run in parallel"
"Error editing file"
"File must be read first"
Repeat 10x for the 10 subagents Claude spawned and then it gets stuck until you press escape and it says "You rejected the parallel agents. Running directly now"
I spent $24,096.47 in "API" costs with my $200 Claude Code Max subscription in April.
I'm building my own saas. I spent 6 months writing the code by hand before using Claude, and that was fine, but its much faster to give the exact specs to Claude and have 3-4 sessions working in parallel with me. When you validate changes with exact test specs there's much less correction you need to do. I always hit my weekly limit and it's far cheaper for me to use this than to hire someone and spend time onboarding them.
I've been on the $200 plan for 3 months, but this will be my last month. I got great use out of 4.5 for a while, but 4.6 felt like a half step back (conflated with all the random hidden config changes during it), and 4.7 is genuinely terrible.
It's impossible to tell these days whether 4.7 is stuck because it's thinking and Anthropic suppressed all output (seriously, 4.7 will just start making changes without explaining any reasoning - how is that an upgrade?) or because the underlying infrastructure is having issues.
4.5 -> 4.7 feels like going from working with a coach-able, junior engineer that does well with clear guidance to working with a cocky mid-level that will spend too long on pointless tangents and make confidently incorrect changes without any discussion.
Lol, spinning up swat teams because someone high up decides "drop everything this is my pet priority now" is politicking. It looks good for the leaders, meanwhile its the engineers pulling the all nighters and dealing with having to maintain systems that are operationally compromised from day 0 because there's no proper planning/scoping involved other than "Big Man says this needs to be done in 2 weeks"
Yeah, I'm sure the numbers are a bit inflated compared to API, but with my Claude $200/month subscription I've supposedly consumed 12,160,410,828 tokens in April for a cost of $22,733.03.
Yes, I throw away maybe 98% of the physical mail I receive. But I have to sort through it for the 2% that's maybe important and for some reason MUST be received by physical mail.
There's obviously a lot of physical waste in this day and age and it's arguably small by comparison, but the paper waste from mail spam still disgusts me.
Resuming from sessions are still broken since Feb (I had to get claude to write a hook to fix that itself), the monitoring tool doesn't work and blocks usage of what does (simple sleep - except it doesn't even block correctly so you just sidestep in more ridiculous ways), and yet there seems to be more annoying activity proxies/spinner wheels (staring into middle distance)... Like I don't know how in a span of a few months you lose such focus on your product goals. Has Anthropic reached that point in their lifecycle already where their product team is no longer staffed by engineers and they have more and more non-technical MBAs joining trying to ride the hype train?
So it sounds like you were using it on auto mode then if it went ahead and fixed the vulnerabilities without additional turns? If so, that isn't really a single prompt.
Can we get a count of people that have had Claude read irrelevant documents or perform unnecessary web searches even when told not to from the beginning?
I'm starting to wonder if this increased token usage is inadvertently bleeding into how Anthropic actually trains their model, especially leading up to IPO. As older models are deprecated and users are forced onto newer models, if the default is less efficient and more token expensive that directly results in higher "profit" for Anthropic in terms of the consumption their users have to tolerate - lest they jump to a competitor.