Anthropic Reviews 115

I use it for writing and like what it produces. Don't like the limits and the customer support is non-existent. I had an issue with the interface and rather than getting an individual agent to resolve it they kept sending me a standard email response. As a paid user of a tech product, I expect access to technical support. Will cancel my subscription and stick with another AI.

Burns up your tokens in no time. Sometimes gives good answers sometimes spins wheels and gives stupid thinking memes while burning your allowance

Charged me twice! And when I disputed the second charge, they downgraded my account. Why would one want to pay twice for the same service. Greed, pure and simple.

While Claude.AI is pretty good at writing code and sometimes even catching things missed in the prompt, the absolutely draconian limitations on their highest tier is unethical and morally wrong. They quick to take your money but then quick to throttle your sessions with Claude.AI. In fact, write now they throttle me so much that I cancelled my subscription cause I can't get anything done.

Anthropic literally shredded millions of books to give us Claude-slop

Warning: "Pro" Plan is a Paid Demo with Broken Support

I am writing this to warn anyone considering the Claude Pro plan. Despite the "Pro" label, the usage limits make it nearly impossible to complete any meaningful project, and the customer service experience is nonexistent.

1. The "Token Trap" & Short Sessions The most frustrating part is how the limits actually work. Anthropic doesn’t tell you that every time you send a message, Claude re-reads the entire conversation history. If you are working on a complex task or a long piece of code, your "limit" isn't 45 messages—it’s more like 5 or 10. Within 20 minutes of actual work, you’ll hit a wall and be locked out for 5 hours. It feels less like a professional tool and more like a paid demo that punishes you for having a long, detailed conversation.

2. Non-Existent Support and Automated Rejections When I realized the service was unusable for my needs, I reached out for a refund. I wasn't even able to speak to a human. I was met with a support bot that immediately rejected my request, citing "eligibility criteria" without any actual nuance or human review. It is a closed loop designed to keep your money while offering zero path to actual assistance.

Final Verdict: Think twice before signing up. The model itself is smart, but the infrastructure and "Pro" limits are a bait-and-switch. If you need a reliable tool for long sessions, this isn't it. You’re better off looking at competitors who don't lock you out of your own paid account after 15 minutes of work.

Randomly ban a normal user without any reason and customer service no reply for a long time, very shit

Credits were retired/cancelled early. No way to talk to a human. AI says tough luck.

I signed up and then when payment is taken and you want to sign in you are prompted to pay again

Since the start of the year, the Claude Pro subscription has become essentially unusable. Toward the end of 2025, Anthropic publicly highlighted a “doubling” of limits—yet shortly afterward, those limits were reduced again. Not back to the previous baseline, but to a level that is noticeably worse. I used to work productively with Pro and rarely hit the cap. Now it’s nearly impossible to complete even a modest project in one go.

Uploading a multi-page PDF is basically a trap: you can burn through a large chunk of your session limit immediately. And if you attempt “Research” when your session limit isn’t close to full, you’re gambling—there’s a good chance the research won’t even finish before you hit the cap.

The only official alternative is upgrading to the Max plan, which costs $100/month. But even there, I’ve seen plenty of user reports saying they reach limits far faster than they used to.

I’m disappointed, because the product used to be reliable. If the limits aren’t restored to at least the pre-promotion level very soon, users will simply move their work—and their money—to other providers.

What happened:
• December 24, 2025: Legitimate annual charge of 180 EUR for Claude Pro Plan annual
• January 12, 2026: During payment method verification (as required by Claude), an additional 180 EUR was charged
• The verification process charged my card despite Claude’s own documentation stating no charge should occur
• The 180 EUR charge does NOT appear in my billing account – it was not reconciled

Why this is a problem:
• I was charged without authorization
• The charge was not applied to my subscription (no extension, no credit)
• My subscription was already paid through December 2026
• Claude Support has not responded substantively in 11 days – only automated bot replies
What I did:
• Contacted support multiple times (Jan 12–23)
• Requested explanation or refund
• Received zero human response

The outcome:
• I filed a chargeback with my bank
• My research shows this is a systematic billing bug affecting many Claude users
• Anthropic/Claude’s support is non-responsive

My recommendation:
Do NOT use Claude Pro if you value customer service. The verification process triggers erroneous charges. Support is broken. This is not a one-time glitch – it’s a pattern.

I regret this experience. The product is good, but the billing and support are unacceptable for a paid service.

There is not one human being available for customer service. Their desktop app has many faults. Not once have they responded to my urgent support requests. I'm reporting them to the Better Business Bureau.

I’ve repeatedly run into basic functionality problems across multiple sessions, including in brand-new chats. These are not isolated incidents but recurring issues that significantly disrupt usability.

For a company of this scale, ongoing failures in core features are unacceptable and show a lack of quality control and responsiveness to user feedback. After experiencing these problems over and over, I’ve lost confidence in the platform’s reliability.

Because of the continued instability and lack of meaningful improvement, I plan to dispute the charges billed to my account. I expected a far more stable and dependable service.

Like:
* Opus 4.5 does some very good writing and cleaning up of notes.
* 50% less lies than GPT.
* 50% less fake positivity than GPT.
* Some good psychological insight.
* Useful coding and instructions.

Dislike:
* The phone app cuts off at 10 minutes of audio with zero warning.
* After cutting off it refuses to send the audio, eating the whole dictation.
* Then the app freezes.
* After about an hour of dication the audio function stops working for 24 hours and claims "can't connect with server" as an excuse for an extreme metering that is not disclosed and does not go away even at the $100. plan which I bought to stop being interrupted with this BS.

God awful company. Their website suggested I use their API access for my project. I did, put in $30 that has magically disappered after using $1.03 of credits.

Complete and utter scam. Do not give this company any money or access to your payment cards.

Anthropic with their ridiculous weekly limit changes on the Max plans as well as secretly dumping down Claude in the background are literally scamming consumers in the United States. This is not a company that you can rely on and they are going to quickly be caught up with by models and companies that can produce something just as good. My theory is that they are trying to cash out. This is a very short term plan and will not result in long-term brand equity.

Claude's excessive caution makes charging these prices completely unacceptable. It's no longer creatively usable after it flattens out at every prompt. It writes as if the most vulnerable people were reading the answers, which is not the case. Charging such prices for this is unacceptable, as the tool therefore offers no real benefit. Why use something that doesn't work? Claude can no longer follow instructions and is obviously rewriting prompts. What's the point? Anthropic, fix it or please shut it down. You are killing human creativity with your artificial limits and bans of any "risk". Treat adults like adults.

Anthropic, you don't have to be prudish, it's unnecessary.

On Dec 24, Anthropic sent us this cheery message:
“To celebrate the Holidays, we’re doubling your usual usage limits from midnight Dec 25 through end of day Dec 31. No strings attached.”

SOUNDS GENEROUS?
Yes, until you realize the “gift” WAS JUST THE OLD PLAN REBRANDED AS A GIFT.

I pay $20/month for Claude Pro Plan. Why? Because Anthropic promised “5x THE USAGE OF THE FREE PLAN”, plus reliable access for creative work.

REALITY?
I get 4 MESSAGES EVERY 5 HOURS — that’s two back-and-forths before I'm locked out again for hours.

LET’S DO THE MATH:
• Pro = 4 messages per 5 hours
• “5x more” = Free = 0.8 messages per 5 hours = 1 message every 6.25 hours
So unless Anthropic invented FRACTIONAL PROMPTS, someone’s LYING about what Pro delivers.

CLAUDE IS NO LONGER THE CLAUDE YOU USED TO KNOW:
To make matters worse, Claude hallucinates, stalls, or derails ~70% of the time on literary work. You try again to clarify — and BOOM! OUT OF USAGE FOR HOURS. PAYING USERS CAN’T EVEN FINISH A TASK.

AND THEIR “HOLIDAY GIFT”?
From Dec 25–31, we got A TASTE OF WHAT PRO USED TO BE before they quietly gutted it. Now they're pretending it was a bonus. That’s not a gift — that’s a GASLIGHT.

SO WE GOT:
• 20% LESS THAN THE FREE PLAN’S BASELINE
• 36% LESS THAN WHAT WE PAID FOR.
That’s not a bonus. That’s a QUIET DOWNGRADE wrapped in holiday sparkle.
We paid for a service, relied on the advertised limits, and structured our creative, professional, or technical work around that access. Then Anthropic quietly throttled us — without notice, without consent — and called it a celebration.

AND NOW?
They’re not fixing it. They’re PUSHING TO UPGRADE to an even more expensive plan — to BUY BACK WHAT WE ALREADY PAID FOR.
It’s not subtle.
It’s not ethical.
Don’t do it.

Pressuring Pro users to upgrade again — to pay more for WHAT WE ALREADY HAD.
This is BAIT-AND-SWITCH.
This is DETRIMENTAL RELIANCE.
This is a company MONETIZING INDUCED FRUSTRATION.

ANTHROPIC:
• STOP SCAMMING YOUR LOYAL USERS.
• YOUR “5X MORE” IS A SCAM.
• YOUR MARKETING IS “BAIT AND SWITCH”.
• YOUR LIMITS ARE PREDATORY.
• YOUR PRODUCT IS BROKEN. FIX IT!!

USERS:
• Cancel Pro.
• Request refunds.
• Vote with your wallet.
• USE OTHER AIs. Claude is completely broken anyway and impossible to work with for serious users and professionals.

They’ll only fix this if enough of us walk away (maybe).

THE SOLUTIONS:
• Use CHATGPT 4.0 legacy for writers. (Don't use 5 and above, not as good).
• Same price as Anthropic's Pro Plan.
• NO RESETS EVER!
• Make it learn your voice, you'll be blown away.

• Use GROK.
• $30/month for a truly superior product.
• RARE RESETS.
• Grok wihch was a poor choice for literary writers a year ago, has gotten spectacular. It can replicate your voice exactly, like Claude used to be able to do before it started hallucinating and gorging himself on tokens.
• Grok has huge token allowance: It can hold several entire novels in one thread.
• Make it learn your voice, you'll be beyond blown away.

BEWARE, SHADY POLICIES AND ACCOUNTS DELETED WITH NO EXPLANATION!
ALL YOU HAVE IN THE ACCOUNT WILL BE LOST AND THEY'LL NOT ALLOW YOU TO EVEN DO A BACKUP.
I'm a professional writer and I was about to complete my new book, I was using Claude for research purposes (e.g. to investigate the scientific aspects of the story), for spell and grammar checking, and to keep track of character's development and plot elements to avoid inconsistencies.
I had uploaded drafts of my book in a project in my anthropic account, and there were many chats where I was discussing with Claude several ideas that I was implementing.
I have never EVER used my account for ANYTHING even remotely against the T&C of anthropic.
There was an instance where I was discussing with Claude a scene of the story where a certain character was enjoying the feeling of killing her enemies. Claude warned me that the chat had been automatically flagged, and after that one day my account suddenly has been DELETED.
There was ZERO information about the reason of the deletion, just a generic "violation of the T&C".
Apart for this occurrence with the flagged chat, I have no idea what can have caused such ridiculous decision, and even that, if someone examined that chat and wasn't able to see that it was a fictional scene despite it being so clear, this means that the safety team is made of unprofessional and stupid people.
Your account, is in the hands of such people who will delete it for no reason.

I complained with the support and they refused to give me any info.
I filed an appeal and it was rejected once again with no information whatsoever about the reason of the deletion.
I asked to allow me to download my chats and the drafts I had uploaded, explaining that it was needed to meet the deadline with the publisher.
They refused.
Two months of creative work, which I trusted was safe in the hands of a supposedly trustworthy company, wasted completely.

Sure, my bad for being too rushed with the creative process and not doing regular backups. I usually do it, but I was late with the deadline and I thought I could just do all in one go.
I would have never imagined that anthropic would act with such unprecedented unprofessionality.

I recommend you to choose other AI
I still enjoy Claude, but I use it in another provider now, where I can also access Gemini, Grok, GPT etc.
I deeply regret having ever trusted anthropic.
NEVER AGAIN!

TL;DR: Great model, but the service infrastructure is catastrophic. Misleading limits, bugged chats, denied refunds after bot-initiated cancellations, and zero access to human support.

The Setup: Broken Promises On December 14, I decided to upgrade to Claude Pro. Before paying, I tested the free version for my daily workflows and ran out of queries quickly. After a 5-hour wait, I explicitly asked the system how long I could discuss similar technical tasks on the Pro plan. I was assured I could work "throughout the entire day." Based on this promise, I paid for a monthly subscription and even added extra funds to my account balance just in case. That was the first deception.

The Performance: Good Output, Short Lived The answers were decent. However, my "unlimited" discussions were capped after roughly 30 minutes. Furthermore, the model (Opus) occasionally committed egregious errors—generating non-compiling code with obvious bugs. I had to waste my limited tokens correcting these hallucinations. It was frustrating, but manageable.

The Technical Collapse Then, the platform started falling apart.

The Glitch: My main workspace froze and stopped generating responses. The support bot advised me to create a new chat and manually migrate my context. I did this, wasting valuable tokens, just to keep working.

The Paywall Trap: When I hit the limit again, I checked the cost to continue. For just 3 simple follow-up queries (no code, no heavy text), I was charged nearly $3.00. I immediately disabled overage charges.

The Breakdown: Shortly after, Claude stopped working completely for me.

The Support Loophole (The Real Scam) Realizing the service was unusable, I asked the support bot on the site if I could cancel and get a refund due to poor service quality.

The Bot: Apologized, confirmed a refund was possible, and canceled my subscription immediately.

The Reality: I was not refunded for the subscription (which I used for less than 48 hours). They only returned the unused wallet balance.

When I followed up via email, a robotic reply stated that a refund was impossible.

I asked to restore my subscription since they refused to return the money. Refused.

I asked to speak to a human agent. Refused.

I went back to the website’s chat support. The bot claimed the refund had "already been processed" (a lie) and automatically closed the chat. I begged for a human operator, but the system refused and suggested I talk to the bot again. It’s a dead end.

The Verdict Claude is a strong model, but that is its only plus.

Useless support bots.

Impossible to reach a human.

Extremely low request limits compared to promises.

High risk of losing money.

For context, I use competitors daily:

Gemini (Advanced/Pro): Slightly less technical in some nuances, but it offers effectively infinite chats during the workday. Zero technical downtime in my long experience.

GitHub Copilot: Access to multiple models and infinite chats. The only downside is the context limit for large texts, but it’s reliable.

Conclusion I’ll end where I started: You cannot work like this. You charge premium prices for a service that has frequent technical failures. You allow your AI agents to promise refunds and cancel user subscriptions on your behalf, but then you hide behind policies to keep the money. You offer no path to human support, leaving users trapped in a loop of automated rejections. To put it bluntly: This is an absolute disgrace.