Sabitlenmiş Tweet
I’ll be publishing a lot more MCP servers this week and in the coming weeks. Follow me to stay updated as soon as they’re released.
English
Moacir Braga
132 posts
@melvynx It is not black and white. It happened to me and to so many people with Max 5x and 20x plan. If it does rain on your city doesn’t mean raining doesn’t exist and who had to face it is and reported it is a lier. It looks like it didn’t happen to everybody.
English
I just use non-stop my Claude code sub all day, didn't even reach 100% of session usage.
People are lying so much on X.
English
@theo It is not overreacting, I have Max 20x and it went from 0 to 100% 5h usage in a few prompts in a single session without even compacting. It took less than 2h to not be capable of working anymore. I lost 50% of my day yesterday. One day before they were also out…
English
I think people are overreacting to this.
It sucks, but they clearly have a real compute shortage and this was the best path they had without cutting your usage massively
Thariq@trq212
To manage growing demand for Claude we're adjusting our 5 hour session limits for free/Pro/Max subs during peak hours. Your weekly limits remain unchanged. During weekdays between 5am–11am PT / 1pm–7pm GMT, you'll move through your 5-hour session limits faster than before.
English
@shaunralston I moved to OpenCode to be AI Provider agnostic in my daily workflow, I was just tired of switching or paying both $200 plan from OpenAI and Anthropic. I subscribed to multiple cheaper AI providers such as Z.ai and MiniMax, working well so far.
English
Anthropic cutting Claude Opus 4.6 rate limits with 0 warning is frustrating 🤬 my $200/month MAX plan rate limited in 2 hours, mid-project. never had that happen with Codex. Any DEVs thinking of dropping MAX but keeping Pro, for ad-hoc use?
English
@robinebers I did what you said, I left. I had Max 20x. Regarding what you said, it is not exactly that. We know they pay for us to use it, but disturbing people’s workflow without being transparent and saying nothing wrong happened, it only affects 2% of the users, etc, that is bad.
English
people need to shut the fuck up
dax is 100% right here
subscribers genuinely say: “how dare you give me less heavily subsidies tokens!!!!”
y’all need to wake the fuck up and understand the true cost of inference
and trust me, i’ve heard it all
people rationalize why they “deserve” the best model and how anthropic’s cost isn’t “the real cost” of the models
BRO THE REAL COST IS WHATEVER THE PRICE IS
if you don’t like it, leave
it’s not a secret that I’m disgusted by Anthropic’s higher-ups/legal team
but people have become god damn delusional
dax@thdxr
man the responses to the new claude max limits are crazy everyones expectations are so out of whack it's kind of embarrassing to get this mad, i'd just be like damn ok i'll cancel were you guys depending on this shit to keep your grandma alive i don't get it
English
@bridgemindai I did the same, I cancelled mine. I tried GPT 5.4 again and it worked perfectly well, but in the end I decided to use OpenCode with OpenRouter and subscribe those “amazing” plans anymore, too much lack of transparency. Bty, I prefer OpenCode TUI to Claude Code.
English
Just cancelled my $200/month Claude Max subscription.
Rate limits lowered dramatically out of nowhere.
529 overloaded errors daily.
Hit my 5 hour limit in under an hour.
Multiple days of outages this week.
Claude Opus 4.6 is the best coding model when it works.
The problem is it hasn't been working.
Switching to Codex with GPT 5.4 as my main driver.
Better rate limits. More reliable.
Actually available when I need it.
I don't want to leave Claude Code.
I've built BridgeMind with it.
But I can't build a company on infrastructure I can't depend on.
Anthropic, fix the reliability.
Fix the rate limits.
I'll come back the day you do.
English
@elvissun What is annoying me are people saying the folks who reached out limits with a few prompts during this peak time are lying or don’t know how to use Claude Code. I have Max 20x and I considered buying another 20x, but I got upset and simply canceled. Still using the remaining days.
English
which happen first?
A) Max 20x becomes $1000/mo
B) Anthropic burns through all its customer goodwill
Thariq@trq212
To manage growing demand for Claude we're adjusting our 5 hour session limits for free/Pro/Max subs during peak hours. Your weekly limits remain unchanged. During weekdays between 5am–11am PT / 1pm–7pm GMT, you'll move through your 5-hour session limits faster than before.
English
@bydylanlamb @0xgib @MonchitoPelu @trq212 @amorriscode It matters. Today when it happened to me I was analyzing the business rules and the api contracts, all markdown files. If we have multiple files, it will try to run sub agents and it burns faster. The only thing people like me are saying it burned insanely fast.
English
@moa_brg @0xgib @MonchitoPelu @trq212 @amorriscode Size of projects doesn’t really matter if I’m using the tokens at the same rate all day.
English
To manage growing demand for Claude we're adjusting our 5 hour session limits for free/Pro/Max subs during peak hours. Your weekly limits remain unchanged.
During weekdays between 5am–11am PT / 1pm–7pm GMT, you'll move through your 5-hour session limits faster than before.
English
@TelebakJ @0xgib @MonchitoPelu @trq212 @amorriscode By the way, why you think nobody provided it? I am pretty sure we people providing it… Tomorrow I will analyze the same issue again and collect as much data as possible to provide to Anthropic*.
English
@moa_brg @0xgib @MonchitoPelu @trq212 @amorriscode Again if it’s happening SOOO many times with SOOO many people getting a single feedback ID will be easy.
English
@jacob_posel Yes, this is exactly what I am concluding as well. OpenAI is also not that much trustworthy, but Anthropic is the second time they completely mess up with my business workflow. The other happenned when they released that post morten, I really had bad time with that degradation
English
Claude is an amazing model
Claude code is an amazing tool
But Anthropic has shown they are willing to lockout users for various reasons
That’s just too much risk to build your entire business on top of their models or products
English
@bydylanlamb @0xgib @MonchitoPelu @trq212 @amorriscode I am wondering if your project is too simple and you are asking small changes or if these issues affected randomly a group of users. A lot of people including me with Max 5x and 20x plans are facing it and a lot like you are not.
English
@moa_brg @0xgib @MonchitoPelu @trq212 @amorriscode Yes. I start using it from 6am to 9pm… I legit haven’t hit any limits.
English
@bydylanlamb @0xgib @MonchitoPelu @trq212 @amorriscode Try it between 5am to 11am PT. Outside this range it works normally exactly like you are saying. I think too many people faced this problem, it is definitely not something wrong I am doing it. I use Claude Code since the first day they released it, before Max plan existed.
English
@moa_brg @0xgib @MonchitoPelu @trq212 @amorriscode I’ve got the 5x plan and almost need hit limits. wtf are you doing lol
English
@ClaudiuDP @trq212 @amorriscode I am stupid, trying to fix it I uninstalled and removed all .claude files from users folder. Otherwise it would be a pleasure to back and analyze the right time and send a feedback. Tomorrow I think it is probably going to happen again; so if it happens I will be ready.
English
English
@SymplicityTV @RafaCrackYT @trq212 What time was it? I think for my case it happened between 8am-9am UTC -3. Right now it is totally normal working as you are describing it and as it always worked.
English
@RafaCrackYT @moa_brg @trq212 Yea that is 100% a bug. Today I had 4 claude code terminals open working on 4 different small projects with subagents, as well as the desktop claude code app working on a massive project that had 5 separate plans, didn’t hit limit. Same plan as you
English
@TelebakJ @0xgib @MonchitoPelu @trq212 @amorriscode Sure, everybody waked up with the unmanaged will of lying about the same problem, all paid users from long time. Bty, I failed on not sending the feedback, but it happened between 8am-9am UTC -3. I tried to talk to them via chat but an AI talks to us, not a human..
English
@moa_brg @0xgib @MonchitoPelu @trq212 @amorriscode Why haven’t you provided a single feedback ID since there is SOOOOOO much proof? I haven’t encountered it either
English
@kr0der What? They nerfed Opus 4.6, 5 doesn’t exist. Anyway, they nerfed by 10x to 20x during peak period. My Max 20x is useless between 5am to 11am PT.
English
@iamgalba @RafaCrackYT @trq212 Wrong, it would affect the context window which wasn’t even 50% usage. Also, that wasn’t my case because it simply works well out 5am - 11am PT. Now it is consuming normally, very similar prompts, 6% usage of my limit. So they have a bug on that period or they cut by 10x or more.
English
@RafaCrackYT @moa_brg @trq212 Check your /context. If you have tons of MCPs there - here is your answer
English
@0xgib @MonchitoPelu @trq212 @amorriscode Impressive you decided to talk about it to a cybersecurity professional, lol. Man, come on, a lot of people get hacked then… I am considering your comment a joke, I am not good on detecting jokes.
English
@moa_brg @MonchitoPelu @trq212 @amorriscode not their fault you're getting hacked bro try adopting some real security practices
English
@BradGroux @AnthropicAI @claudeai @DarioAmodei @FTC I am backing to OpenAI. So sad, Opus 4.6 and Claude Code are the state of the art, but it became impossible to work even with my Max 20x, not surviving for a single session. They reduced multiple times the capacity of the plans.
English
You make this announcement BEFORE you make the change. Good riddance... what a shame @AnthropicAI and @claudeai. Customer service isn't hard, @DarioAmodei. This was a rug pool and y'all got caught.
I hope the @FTC fines you.
Thariq@trq212
To manage growing demand for Claude we're adjusting our 5 hour session limits for free/Pro/Max subs during peak hours. Your weekly limits remain unchanged. During weekdays between 5am–11am PT / 1pm–7pm GMT, you'll move through your 5-hour session limits faster than before.
English
@RafaCrackYT @claudeai I just noticed out of the time @trq212 said, it works normally. So the bug seems to happen on that time. My simple prompts are not consuming normally, similar prompts are now in 2% usage, but at this morney it went to 100% super fast, so they kind of blocked Max on rush time
English
@MonchitoPelu @trq212 @amorriscode A lot of people complaining about exactly the same problem, a lot of them with Max plan and the guy ask for proof…
English
English