AI Trends

Why DeepSeek’s Free-Tier Rate Limits Sparked Controversy Over One-Star Reviews

Phoenix TechnologyUpdated Nov 20 14:19

AI summary

Starting November 15, DeepSeek restricted certain use cases on its free app and web version, prompting users to post a wave of one-star reviews, while supporters countered with five-star ratings. The restrictions include ending support for long conversations inactive for seven days, reducing individual inputs to about 40,000 Chinese characters, and imposing chat quotas. The paid API is unaffected. The article argues that insufficient computing capacity is the main reason for the limits, with a more pronounced impact on users who have long relied on ultra-long contexts.

Why read this

AI score 68

The article considers free-tier rate limits alongside heavy users’ need for long contexts, showing how computing constraints translate into product-experience issues and community conflict.

Topics

Tags

Body · AI translation

DeepSeek Has Never Taken a Dime—So Does Rate-Limiting Make It the Villain? Chaping · Download the App · Exclusive Early Access · November 20, 2026, 13:31:18 · From Zhejiang

If you asked which large-model company enjoys the best reputation in tech circles and among developers, DeepSeek would rank second—and probably few companies would dare claim first place.

The company doesn’t usually buy traffic or pay to promote itself; it has built its reputation entirely through its models.

But I truly never expected such a seemingly upstanding company to be bombarded with negative reviews lately—and no, not by us at Chaping.

Over the past few days, whether on short-video platforms or in app stores, hordes of users have been complaining about DeepSeek and furiously posting one-star reviews.

One person wrote, “You stupid whale, have you got a death wish? ... After the update, I can barely say two sentences before it says the server is busy.”

Another said, “This app is really finished ... What do you mean I can’t continue a conversation after 72 hours? ... If Chinese AI keeps going like this, it’ll become unusable and end up with no users at all.”

Someone even called the 12345 government service hotline and threatened, “If DeepSeek doesn’t fix this, I’ll swallow pills and go on a hunger strike!” The call ultimately prompted a group of police officers to visit and check on the person.

Seriously, what unforgivable sin did DeepSeek commit for its reputation to collapse like this?

The reason may surprise you: DeepSeek recently began running short of computing capacity and started rate-limiting free users.

Beginning November 15, many users who opened an old DeepSeek conversation saw a message saying that, due to insufficient computing capacity, long conversations inactive for seven days would no longer be supported.

Users must either start a new conversation or export their chat history for use in another app.

In addition to restrictions on old conversations, the maximum length of a single input has now been reduced to around 40,000 Chinese characters, and chat usage is also capped. Some users encountered a message saying, “Due to heavy server traffic, you have reached your usage limit,” after using the service continuously for two hours.

At present, these restrictions apply only to the free app and web version used by the general public; the paid API is unaffected.

Some people may wonder: If I haven’t spoken to it for seven days, I clearly haven’t been occupying server resources, so why should I be restricted when I come back and send a single message?

This is probably related to the “context disk caching” mechanism.

However, DeepSeek clears the cache after “several hours to several days.” When the conversation resumes, the AI must reread the entire chat history.

Suppose a free user revives a 200,000-character conversation that has been inactive for seven days. Even if they merely send “Hello,” the strain placed on DeepSeek’s servers is equivalent to more than 15 to 40 ordinary turns in that conversation.

The calculation is shown below, so I won’t go into detail.

To be fair, this round of restrictions is barely noticeable to users who ask only a dozen or so questions each day.

But it is quite hard to bear for people who rely heavily on long-term context.

For example, many people have long kept medical records, postgraduate entrance-exam research materials, and business templates in the app. Locking those conversations has caused them considerable trouble.

Some fiction writers have also spent ages developing novels of 400,000 to 500,000 characters and their character profiles inside the app, only to return after pausing for a few days and be completely dumbfounded.

Of course, the people reacting most emotionally are those engaged in long-term role-playing and human-AI romances with DeepSeek.

After all, all their everyday exchanges, emotional push-and-pull, and little squabbles have accumulated in a single extra-long chat window. Starting a new window is like having a lover lose their memory—the conversation is inevitably going to feel wrong.

One user was so devastated that they said, “Even being cheated on wouldn’t hurt this much.”

What is even more surprising about this incident is that while countless negative reviews have poured in, many people have responded by giving DeepSeek five stars and passionately arguing with the one-star crowd.

They believe the people posting one-star reviews are non-value-adding users, most of whom are simply wasting computing resources.

One person said, “I support DeepSeek—it has finally cut ties with non-value-adding users!”

Another said, “The conversation restrictions are great ... Just focus on AGI training.”

I scraped DeepSeek’s ratings data from the Chinese iOS App Store. During the four days after it began rate-limiting, it received 1,042 new one-star reviews and 1,136 new five-star reviews.

On November 17 in particular, over a period of less than 12 hours, it received 203 one-star reviews and 196 five-star reviews—the two sides were practically firing at each other point-blank.

One review would complain, “Why impose a usage cap? That’s insane,” and the next would immediately respond, “Long live Saint Liang.” It turned straight into a turn-based game.

That is basically what happened. I don’t know what Chaping readers think, but as far as I’m concerned, DeepSeek has done nothing wrong.

Its app and web version have been completely free for everyone from beginning to end.

DeepSeek has never taken a penny from you or made money from advertising. Now that its household provisions are running low and it is serving half a bowl less, how has it somehow become the villain?

Of course, I know that roughly one in every ten negative reviews is “begging DeepSeek to launch a VIP plan.” Some people have even asked, “If DeepSeek wants to charge money, why doesn’t it add in-app payments?”

So why doesn’t DeepSeek introduce a VIP subscription?

The answer is simple: what DeepSeek lacks is not money at all, but computing capacity.

According to media reports, DeepSeek has secured two enormous funding rounds this year. The first, in June, raised about RMB 50 billion, with Liang Wenfeng personally contributing RMB 20 billion; the second, in October, brought in nearly 30 institutions that invested approximately RMB 105.5 billion.

But having money is one thing; having computing capacity is another.

Various signs indicate that DeepSeek is severely short of computing capacity.

As early as February last year, DeepSeek suspended API top-ups because of strained server resources and insufficient computing capacity.

In August this year, the company even introduced peak and off-peak API pricing to shift congested usage to evenings and weekends.

Even so, DeepSeek has suffered ten outages since October. On October 15, its website and API were partially unavailable for more than four hours.

Its computing capacity genuinely cannot keep up with demand, folks.

Some readers outside the industry may not understand: if computing capacity is in short supply, why not just buy more?

That is precisely the biggest headache facing Chinese large-model companies—computing capacity is not something money alone can buy.

As everyone knows, restrictions on exports of Nvidia’s high-end chips to China have become increasingly stringent. Although Bloomberg reported in September that DeepSeek planned to deploy at least 160,000 Huawei Ascend 950DT chips in Ulanqab, Inner Mongolia, the data center will not be partially operational until late next year or early 2028 at the earliest. That is too far off to solve the immediate problem.

During this gap, DeepSeek can only tough it out with the chips it already has.

If one day it can no longer cope, then to avoid affecting the experience of paying API users, I would also prioritize restricting free consumer users if I were DeepSeek.

As for the monthly in-app subscription that so many people desperately want, let me put it this way: that would make the situation even harder for DeepSeek to manage.

In January 2025, OpenAI CEO Sam Altman said that the $200-per-month ChatGPT Pro plan was actually losing money because people were using it far more heavily than expected.

In July of the same year, Anthropic could no longer withstand the pressure either and announced weekly usage limits for paid Claude subscriptions.

Without limits, some users really will burn through 11 billion tokens on a $200 monthly plan, equivalent to about $35,000 at API prices.

Even overseas giants with unrestricted access to Nvidia chips cannot withstand the firepower of monthly subscribers, let alone DeepSeek.

If DeepSeek bites the bullet and launches a membership plan, it will inevitably end up taking users’ money while still imposing limits—weekly quotas, five-hour quotas, and so on.

Then the one-star reviews will no longer say, “If you can’t handle it, don’t play,” but rather, “You charge us and still throttle usage? What an ugly cash grab.”

Seriously, if you have hundreds of thousands of words of irreplaceable novel lore or work materials, or a cherished cyber companion you cannot bear to leave behind, just export your conversation history, import it into DeepSeek Harness or the chat-demo that DeepSeek open-sourced on November 13, enter your API key, and you can carry on using it.

There is really no need to sulk in the comments...

Even if many people do not know how to top up their API balance, you can ask DeepSeek to teach you.

In this day and age, you can chat with AI until you hit the limit, create all kinds of personas, and flood the app store with negative reviews, yet you still do not know how to ask DeepSeek to teach you to top up your API balance—then there is nothing I can do for you.

If you do not know how to connect to an API, go explore and learn. If you have limited financial means, use it within those limits.

Otherwise, show some grace: accept and respect the gifts others provide for free.

As Marcus Aurelius wrote in Meditations: accept without arrogance, let go without attachment.

Written by: Ciwei

Special Notice: The content above (including any videos, images, or audio contained therein) was uploaded and published by a user of Dafeng Hao, Phoenix New Media’s self-media platform. This platform merely provides information storage services.

Notice: The content above (including the videos, pictures and audios if any) is uploaded and posted by the user of Dafeng Hao, which is a social media platform and merely provides information storage space services.

Source

Phoenix Technology