your usage patterns may have been similar to usage patterns of other customers that are often problematic, so some automated anti-abuse system made some statistical decision & banned you. you're a false positive, it's not worth their time reviewing the decision as most people they ban are problematic customers, and they've estimated that you're probably problematic.
Perhaps the cost of providing a 20x subscription exceeds the monthly subscription revenue, if a customer uses it heavily. Perhaps most heavy users only have a single subscription, and they can account for running at a small loss as a marketing expense & hoping you may convert to a profitable tier in future. But if you have 13 subscriptions of their 20x consumer plan at once, maybe they lose so much money from it that they decide it's not worth continuing to doing business with you -- purely commercially -- so at their discretion they ban you.
In both cases, they've decided it isn't worth doing business with you, for different reasons, so if they estimate that a new customer signing up is in fact you, the same person they previously decided to ban, they'll keep banning your attempts to open new accounts. They're not going to explain how they figure out you're the same person as that undermines the effectiveness of their banning process.
If you were banned purely because you're costing them too much money vs what you are paying, in theory that's fixable if you are willing to pay enough money to sign up for a proper business plan & can find some salesperson to negotiate with. but in practice presumably they have all their salesfolk pointed at enterprise customers because that is where the real money is.
Who knows, I’m not surprised that I got banned. I probably looked like an api reseller or some bot farming network. I’m sure my cost was higher than my subscription fees, and that was probably enough.
Just out of curiosity: How did you mentally keep oversight of so many terminal doing things all around the place? Or is this done without too much oversight and do you just look at the results like sales volumes?
The terminals get a red background tint when they are running, and turn green when they are finished a task, amber if they need a response.
I’d always ask for a tldr, then just do some brief testing. Most of my prompts would run for an hour or more, so it was really just checking & responding to a few windows every hour while I do my other work.
It was always small tasks, given way too much compute, just so I could always be confident in the result. I always gave strict requirements, and if I was really lazy I’d say “iterate over this 10 times once finished, test everything, only come back to me with a blocker or if you have an extremely high level of certainty” some crap like that
It was a really lazy way to use AI, but it just worked well for me for the small tasks I needed to do, knew how, but wanted them automated, or if it was a once off - I just couldn’t be fucked, or busy with something else so I’d just ask a new terminal to do it.
Eg. My supplier is asking for sales forecast, fetch my sales from Shopify, look at seasonality from my stock-app and build a report and email to xxx@mysupplier.com - triple check it against Shopify, xero, report any anomalies
Then just blast that out on fable ultra code and it would rarely mess things like that up… I mean opus would hardly mess things like that up, but I’m just so short on time that I palm off any task I can, even tasks that might take just as long telling Claude but it’s much funner getting Claude to do it than repeating some process I’ve done a thousand times
Oh and I’m using codex in the meanwhile, it’s pretty good, but just finding its overall planning to not be as good, and it seems to have a tendency to over engineer. Maybe it’s just prompting I need to change, but I’d really like just one account to do planning with Fable, then pass to codex.
(Also, before people say “just pay API pricing” - it’s not worth it for how much money I am earning. $2600usd/month I know seems like a lot, but if I switched to API usage, it would be 10x that.
I import physical products from China and sell them online in Australia. Would prefer not to disclose my niche, but it’s the equivalent of selling bedding/pillows. Nothing at all unique or interesting, the products are good and well priced though.
Are you running the direct data of your workload all through the AI? The number of tokens you're consuming can probably be reduced a lot. It seems like a good time to take a half step back and use the AI to build a workflow engine to shift more of the work out of the AI models and into more traditional scrape and analyze sw logic. For the more sophisticated decisions, even using a LLM grade AI to bootstrap more older machine learning algorithms and rules could AI use to updates of rules instead of something called inline on every working transaction.
Get a computer with a good graphics card and try out opencode. The quality has recently improved a lot (comparable to Claude about 12 months ago) and everything is running locally, so you'd save on costs but most important: nobody can kick you out.
Author wrote that opus required too much handholding, to host opus locally (aka glm 5.2) you need $50k in hardware (or $3k/mo to rent a slow server) or just buy tokens from hoster.
After the rejected appeal, cycling VPNs and credit cards probably looks like ban evasion, whatever triggered the original ban. I’d stop creating accounts and ask Anthropic to clarify who is actually barred.
Would love to ask anthropic, but there is no way of contacting them. Once they reject your appeal, there is no option. No phone number, no email, and can’t access live chat without a paid account.
Couple of guesses:
your usage patterns may have been similar to usage patterns of other customers that are often problematic, so some automated anti-abuse system made some statistical decision & banned you. you're a false positive, it's not worth their time reviewing the decision as most people they ban are problematic customers, and they've estimated that you're probably problematic.
Perhaps the cost of providing a 20x subscription exceeds the monthly subscription revenue, if a customer uses it heavily. Perhaps most heavy users only have a single subscription, and they can account for running at a small loss as a marketing expense & hoping you may convert to a profitable tier in future. But if you have 13 subscriptions of their 20x consumer plan at once, maybe they lose so much money from it that they decide it's not worth continuing to doing business with you -- purely commercially -- so at their discretion they ban you.
In both cases, they've decided it isn't worth doing business with you, for different reasons, so if they estimate that a new customer signing up is in fact you, the same person they previously decided to ban, they'll keep banning your attempts to open new accounts. They're not going to explain how they figure out you're the same person as that undermines the effectiveness of their banning process.
If you were banned purely because you're costing them too much money vs what you are paying, in theory that's fixable if you are willing to pay enough money to sign up for a proper business plan & can find some salesperson to negotiate with. but in practice presumably they have all their salesfolk pointed at enterprise customers because that is where the real money is.
Who knows, I’m not surprised that I got banned. I probably looked like an api reseller or some bot farming network. I’m sure my cost was higher than my subscription fees, and that was probably enough.
Change computers or operating systems as well as ips. These days they get info from the OS and will ban you based on your profile.
Just out of curiosity: How did you mentally keep oversight of so many terminal doing things all around the place? Or is this done without too much oversight and do you just look at the results like sales volumes?
The terminals get a red background tint when they are running, and turn green when they are finished a task, amber if they need a response.
I’d always ask for a tldr, then just do some brief testing. Most of my prompts would run for an hour or more, so it was really just checking & responding to a few windows every hour while I do my other work.
It was always small tasks, given way too much compute, just so I could always be confident in the result. I always gave strict requirements, and if I was really lazy I’d say “iterate over this 10 times once finished, test everything, only come back to me with a blocker or if you have an extremely high level of certainty” some crap like that
It was a really lazy way to use AI, but it just worked well for me for the small tasks I needed to do, knew how, but wanted them automated, or if it was a once off - I just couldn’t be fucked, or busy with something else so I’d just ask a new terminal to do it.
Eg. My supplier is asking for sales forecast, fetch my sales from Shopify, look at seasonality from my stock-app and build a report and email to xxx@mysupplier.com - triple check it against Shopify, xero, report any anomalies
Then just blast that out on fable ultra code and it would rarely mess things like that up… I mean opus would hardly mess things like that up, but I’m just so short on time that I palm off any task I can, even tasks that might take just as long telling Claude but it’s much funner getting Claude to do it than repeating some process I’ve done a thousand times
> The terminals get a red background tint when they are running, and turn green when they are finished a task, amber if they need a response
You can probably reach higher productivity by having the computer reward you with a banana after responding
Oh and I’m using codex in the meanwhile, it’s pretty good, but just finding its overall planning to not be as good, and it seems to have a tendency to over engineer. Maybe it’s just prompting I need to change, but I’d really like just one account to do planning with Fable, then pass to codex.
(Also, before people say “just pay API pricing” - it’s not worth it for how much money I am earning. $2600usd/month I know seems like a lot, but if I switched to API usage, it would be 10x that.
What's your business?
I import physical products from China and sell them online in Australia. Would prefer not to disclose my niche, but it’s the equivalent of selling bedding/pillows. Nothing at all unique or interesting, the products are good and well priced though.
Are you running the direct data of your workload all through the AI? The number of tokens you're consuming can probably be reduced a lot. It seems like a good time to take a half step back and use the AI to build a workflow engine to shift more of the work out of the AI models and into more traditional scrape and analyze sw logic. For the more sophisticated decisions, even using a LLM grade AI to bootstrap more older machine learning algorithms and rules could AI use to updates of rules instead of something called inline on every working transaction.
Get a computer with a good graphics card and try out opencode. The quality has recently improved a lot (comparable to Claude about 12 months ago) and everything is running locally, so you'd save on costs but most important: nobody can kick you out.
Author wrote that opus required too much handholding, to host opus locally (aka glm 5.2) you need $50k in hardware (or $3k/mo to rent a slow server) or just buy tokens from hoster.
After the rejected appeal, cycling VPNs and credit cards probably looks like ban evasion, whatever triggered the original ban. I’d stop creating accounts and ask Anthropic to clarify who is actually barred.
Would love to ask anthropic, but there is no way of contacting them. Once they reject your appeal, there is no option. No phone number, no email, and can’t access live chat without a paid account.