Yep, i have long felt like i randomly get sonnet results despite opus billing. I try to work odd hours and notice better results.
Ask HN: Claude Opus performance affected by time of day?
11–20 of 45 posts
Re: Ask HN: Claude Opus performance affected by time of day?
#12For what it’s worth, Anthropic very strongly claim that they don’t degrade model performance by time of day [1]. I have no reason to doubt that, imo Anthropic are about as ethical as LLM companies get. [1] https://www.anthropic.com/engineering/a-postmortem-of-three-...
Banning paying users with no warning doesn’t seem super ethical. Probably not unethical, either, but I would not frame them as “the most ethical”
Re: Ask HN: Claude Opus performance affected by time of day?
#13[flagged]
Re: Ask HN: Claude Opus performance affected by time of day?
#14It seems clear that, rather than throttling, anthropic serves lower quality versions of their models during peak usage to keep up with demand. They refuse to admit it, and it's hard to prove, but these threads consistently happen ~3 months after every single model release.
Re: Ask HN: Claude Opus performance affected by time of day?
#15It’s possible that they could be using fallback models during peak load times (west coast mid day). I assume your traffic would be routed to an east coast data center though. But secretly routing traffic to a worse model is a bit shady so I’d want some concrete numbers to quantify worse performance.
To be clear, the company has very directly denied doing this.
Re: Ask HN: Claude Opus performance affected by time of day?
#16For what it’s worth, Anthropic very strongly claim that they don’t degrade model performance by time of day [1]. I have no reason to doubt that, imo Anthropic are about as ethical as LLM companies get. [1] https://www.anthropic.com/engineering/a-postmortem-of-three-...
Banning paying users with no warning doesn’t seem super ethical. Probably not unethical, either, but I would not frame them as “the most ethical”
I'd say they're about as good as the average billion dollar American tech company when it comes to ethics.
Re: Ask HN: Claude Opus performance affected by time of day?
#17Earlier quoted context omitted.
I suspect they might transparently fall back too; Opus 4.5 has been really reasonable lately, except right after it launched, and also surrounding any service interruptions / problems reported on status.claude.ai -- once those issues resolve, for a few hours the results feel very "Sonnet", and it starts making a lot more mistakes. When that happens, I'll usually just pause Claude and prompt Codex and Gemini with the…
Precisely. Once I point out the fact that it is doing this, it seems to produce better results for a bit before going back to the same. I jokingly (and not so) thought that it was trained on data that made it think it should be tired at the end of the day. But it is happening daily and at night.
I find it helps to tell it to take some stimulants
Re: Ask HN: Claude Opus performance affected by time of day?
#18Are you using the API or a subscription?
Re: Ask HN: Claude Opus performance affected by time of day?
#19I had something similar with GPT, like a clockwork every day after like 1pm it started producing total garbage. Not sure if our account was A/B tested or they just routed us to some brutal quantization of GPT, or even a completely different model.
Re: Ask HN: Claude Opus performance affected by time of day?
#20I mostly use Gemini, so I can't speak for Claude, but Gemini definitely has variable quality at different times, though I've never bothered to try to find a specific time-of-day pattern to it. The most reliable time to see it fall apart is when Google makes a public announcement that is likely to cause a sudden influx of people using it. And there are multiple levels of failure, first you start seeing iffy responses…
I suspect they might transparently fall back too; Opus 4.5 has been really reasonable lately, except right after it launched, and also surrounding any service interruptions / problems reported on status.claude.ai -- once those issues resolve, for a few hours the results feel very "Sonnet", and it starts making a lot more mistakes. When that happens, I'll usually just pause Claude and prompt Codex and Gemini with the…
I didn't believe such conspiracy theories, until one day I noticed Sonnet 4.5 (which I had been using for weeks to great success) perform much worse, very visibly so. A few hours later, Opus 4.5 was released.
Now I don't know what to think.