New blog posts have been thin on the ground here lately. The BuchhalterPython series has stalled. Both of those things have the same cause: Anthropic quietly made Claude Pro nearly useless for me, then told me that was the plan.
What I Actually Use
Claude Pro. €20 a month. Sonnet 4.6 only. No Opus. No 1M context sessions. I am the most basic paying customer they have.
In February I started using the paid plan. I had no illusions — €20 is not a lot of money. Hitting session limits when working across three projects at once was expected. But I could work for three to four hours before a limit kicked in. That was enough to get real things done.
In March there was a double-token promotion. Everything worked well. I built infrastructure, provisioned containers, wrote code, generated documentation. The posts from that period are on this site if you want to read them.
Then It Stopped Working
About a week ago, limits started hitting in 20 minutes. Not three or four hours. Twenty minutes. The weekly limit was gone within three days. Nothing changed on my end.
A quick search turned up the explanation. Anthropic introduced “Peak Time” limits without telling anyone. Peak Time is 5:00 am to 11:00 am Pacific. For me, a hobbyist in Central Europe, that is 14:00 to 20:00 CET. The hours when I actually sit down for hobby projects like this one.
So: peak limits exactly during my hobby hours, tighter by design. Good to know.
Off-peak is not much better. I watch the token usage bar constantly now. Just loading a single issue into context takes several percentage points off the daily allowance. I cannot finish one task in a flow. I cannot generate a post. I cannot write documentation from start to finish without hitting a wall.
It is back to Gemini in chat mode with copy and paste. Manually. Like 2023.
Reddit Was Already on Fire
I am not alone. Reddit has been full of complaints for days: no transparency, no communication, no warning. Just suddenly hitting limits that used to take hours in a matter of minutes.
Three days ago Anthropic posted on Reddit that they were “investigating”. That post attracted hundreds of comments from users saying exactly what I am saying.
Then, a short time ago, the follow-up appeared. Official response from Anthropic:
“Peak-hour limits are tighter and 1M-context sessions got bigger, that’s most of what you’re feeling. We fixed a few bugs along the way, but none were over-charging you.”
Peak limits are tighter. The big context window got bigger, so the budget went somewhere else. Working as intended. The rest of the post is tips — don’t use Opus, avoid large context windows, watch your prompt cache — things every normal user already does. I certainly do. Sorry you feel that way.
That Is Gaslighting
They changed the product. They did not announce it. Users noticed and complained. Anthropic waited several days, then confirmed the change was deliberate while implying the problem is that users do not understand how tokens work.
I do not use Opus. I do not use the 1M context window. The “fixes” they mention do not apply to me. My Pro account behaves like the free tier did six months ago.
I have not had my subscription long. Switching to another provider is not a big lift technically. I actually like Claude’s models — the quality is real. But a product that changes without notice, and a company that responds to user complaints by explaining that the product is working correctly, is not one I can plan around.
I am watching how the fallout develops before I decide what to do next.