Why is this front page news here?
Because people upvoted it.
Elevated Errors for Opus 5
71–80 of 96 posts
Re: Elevated Errors for Opus 5
#72Earlier quoted context omitted.
Over the last couple months, I’ve gotten the impression that OAI does indeed have the better infra setup and team, not just more capacity. I don’t think Codex’s 99.98% uptime vs Claude Code’s 99.44%, is just due to OAI having more hardware to run it on. https://status.openai.com/ https://status.claude.com/
Having more servers does help more than you think though. us-east-1 is the AWS region with the most downtime by a large margin simply because it’s the region with the highest demand. Having too much demand on limited servers means that less of those servers are going to be available for redundancy.
Isn't us-east-1 also an early rollout region for new stuff as well?
Re: Elevated Errors for Opus 5
#73Maybe optimize the harness for less turns and less tokens to deliver real value as opposed to the turn taking token hungry so you can less load on your servers, oh, right, your entire bottom line is tokens.
I don't understand this take. Ultimately, people pay for how much they're able to accomplish with the model, not raw token count. If Anthropic thought people could accomplish the same amount with fewer tokens, they'd adapt the harness to do that and then raise the cost of tokens to make more profit (or lose less).
Re: Elevated Errors for Opus 5
#74Earlier quoted context omitted.
I have programmed since I was 5, so 46 years now. I have no problem doing it, and still do. I however wouldn't waste my time only coding by hand when an LLM can do it far faster.
Nobody said "only by hand", you moved the goalpost. If you don't practice it you lose the skill.
I have not moved any goal posts. You made unwarranted and frankly grossly rude assumptions.
Re: Elevated Errors for Opus 5
#75[dead]
Re: Elevated Errors for Opus 5
#76[flagged]
Re: Elevated Errors for Opus 5
#77Earlier quoted context omitted.
Having more servers does help more than you think though. us-east-1 is the AWS region with the most downtime by a large margin simply because it’s the region with the highest demand. Having too much demand on limited servers means that less of those servers are going to be available for redundancy.
Isn't us-east-1 also an early rollout region for new stuff as well?
When I worked there no, we wouldn’t test in the most popular region first.
Re: Elevated Errors for Opus 5
#78Earlier quoted context omitted.
Nobody said "only by hand", you moved the goalpost. If you don't practice it you lose the skill.
I have not moved any goal posts. You made unwarranted and frankly grossly rude assumptions.
So you think if you don't practice you don't lose the ability, that's fine
Re: Elevated Errors for Opus 5
#79And unlike OpenAI, Anthropic don't seem to reset weekly usage quotas after an outage. Anthropic seem to be both letting their competitors outplay them and making unforced errors (like the anti Open Weights models stuff and the frequent Fable->Opus downgrades for 'safety'). Outages are inevitable in this early high growth, rapid development era, but tactical errors are not.
Anthropic would give out resets like OpenAI if they had capacity to do so, it's not like they haven't thought about doing the obvious. OAI waits until overall cluster demand goes down before giving out a reset, they just happen to have much more spare capacity than Anthropic.
Source?
> it's not like they haven't thought about doing the obvious.
Re: Elevated Errors for Opus 5
#80Maybe optimize the harness for less turns and less tokens to deliver real value as opposed to the turn taking token hungry so you can less load on your servers, oh, right, your entire bottom line is tokens.
I don't understand this take. Ultimately, people pay for how much they're able to accomplish with the model, not raw token count. If Anthropic thought people could accomplish the same amount with fewer tokens, they'd adapt the harness to do that and then raise the cost of tokens to make more profit (or lose less).
I mean, it's a near daily experience for me that Claude tries to 'sudo pacman -S' some dependency multiple times before giving up and admitting that it's fundamentally impossible for it to run that command in the first place.
I could accomplish a lot more with those tokens if it just asked for the package to be installed.