Earning 20,000 yuan a month still can't afford a "lobster"? Why is "lobster" so expensive?

Investing in stocks? Rely on JinQilin Analyst Reports—authoritative, professional, timely, comprehensive—helping you discover potential thematic opportunities!

(Source: Titanium Media APP)

Recently, almost the entire internet has been buzzing about OpenClaw lobster farming, but among the praise, there are some discordant voices. The most talked-about is the news that “earning 20,000 yuan a month can’t support lobster farming.” Why are lobsters so expensive?

1. Can’t support lobster farming with a 20,000 yuan monthly salary?

According to Wall Street Insights, someone posted a “Lobster Farming Diary” in their social circle. Nearly a thousand people queued outside Tencent Tower, and some were willing to pay 500 yuan for on-site installation, eager to get it running quickly.

Many didn’t expect that the cost of OpenClaw isn’t in the software itself but in the underlying model calls. It’s inherently a “Token black hole”: each task consumes a large amount of tokens interacting with the backend large model. As task chains lengthen, tool calls increase, and memory is enabled, consumption rapidly rises.

A regular chatbot might use only a few hundred tokens for a round trip; but OpenClaw performing the same task could require millions of tokens. Users report that searching for information or writing a 2,000-word document can burn through 7 million tokens; running a simple crawler test can consume 29 million tokens; cases of daily token consumption reaching 50 million are common.

One SaaS company even provides lobster subsidies for all employees—ordinary staff spend about 150 yuan daily on tokens, while the tech team spends up to 1,000 yuan. More covertly, OpenClaw’s built-in “heartbeat mechanism” consumes about 145 yuan worth of calls daily, even without actual output, totaling over 5,000 yuan monthly.

Some media outlets even lament, “Why do people keep saying these days: earning 20,000 yuan a month can’t support OpenClaw? Actually, that’s true. If you want to use it comfortably, 20,000 yuan isn’t enough to burn; you can only experience it on a tight budget.”

2. Why are lobsters so expensive?

As an open-source software, OpenClaw should have gained wider adoption and popularity through its open nature. However, the reality is that using it is like raising an expensive pet—despite a decent salary, it can still feel burdensome. So, what causes its costs to be so high?

First, open source doesn’t mean free. The hidden costs of OpenClaw add up to form the basic framework of its usage expenses. Many users are attracted by its open-source nature, mistakenly believing they can enjoy its powerful features at zero cost, but they overlook the core logic of open-source software: open source is “code openness,” not “free usage.” The value of open-source software lies in lowering development barriers, not eliminating all costs. From deployment, costs begin to accumulate and tend to be “hidden and ongoing.”

For individual users, lightweight deployment on existing computers is possible without extra hardware purchases, but setting up the compatible environment involves debugging and configuring various open-source tools. Without professional technical skills, additional technical service fees are needed. For enterprise users, to ensure stability and concurrency, they must purchase dedicated servers, set up clusters, or even configure high-performance GPUs—these hardware investments often cost tens of thousands of yuan, representing a significant one-time expense.

Additionally, OpenClaw’s daily operation depends on various third-party services and plugins—whether for speech synthesis, web crawling, or messaging platform integration—some core functionalities require paid services. These hidden costs seem small individually but accumulate over time, significantly increasing overall usage costs.

Second, the cost of calling large model APIs is the core expense of daily OpenClaw use. Its core capability relies on the inference power of backend large models. Each interaction, decision, or code generation consumes tokens in real time. According to current media reports, even for lightweight personal use—such as daily Q&A, simple document organization, or basic email replies—monthly token consumption easily falls between 1 million and 3 million, roughly 20 to 80 yuan.

This may seem negligible, but at high-frequency automation scenarios, costs grow exponentially. When users process large batches of files, run multi-task agents, or perform large-scale web crawling, monthly token consumption can soar to 3 million to 10 million or more, with corresponding costs reaching 80 to 300 yuan or higher.

For businesses or heavy individual users relying on automation, this seemingly small “traffic fee” can, when scaled up, swallow most of the human labor savings. Tokens are no longer just technical metrics but have become “oil” in the digital economy era—price fluctuations directly impact automation applications. For most large model companies, this has become a highly profitable path. For example, the “Dark Side of the Moon” project earned over what is projected for the entire 2025 year within just 20 days, signaling that overseas revenue has for the first time surpassed domestic. In response to the booming OpenClaw, the company quickly launched Kimi Claw, a cloud agent product priced above 199 yuan for paid users, becoming the first among the “Five Little Tigers” in China to develop a cloud-based agent. This shows how lucrative token-based business can be.

3. Frequent automated tasks further increase token consumption invisibly. OpenClaw’s boast of automation hides a lot of “invisible work”—these silent computational costs are the hidden killers of high expenses. Many users’ understanding of costs is limited to “what results I see,” ignoring “what the machine did behind the scenes.” The core value of OpenClaw lies in its agent capabilities—automating model handling, task planning, and self-iteration. But these are also the most prone to cost overruns.

Imagine executing a seemingly simple task like “organize meeting minutes and extract to-do items.” It doesn’t just give an answer directly. Behind the scenes, it needs to convert speech to text, call the large model for semantic analysis, format the output, and possibly perform self-reflection to verify accuracy. Each step consumes大量 tokens. These “backstage efforts” are invisible to users but are the root cause of cost surges.

More critically, to ensure stable automation, OpenClaw often requires extensive trial and error. The model may perform multiple internal reasoning and self-corrections before producing the final answer. Unlike traditional software, where running a piece of code consumes fixed computing power, AI agents like OpenClaw may experience exponential token consumption due to context misunderstandings or randomness.

This uncertainty cost is unprecedented in traditional software engineering. Users may think they just made the “lobster” crawl a little further, but internally, countless “mental calculations” are happening—each deducting tokens. This “black box” cost mechanism is the fundamental reason why users feel they’re spending a lot of money for little apparent work.

4. Tokens are increasingly becoming a structural bottleneck restricting OpenClaw’s popularity. From an economic perspective, the widespread adoption of a technology depends on whether its marginal cost is below the human labor cost it replaces. Currently, the high price of tokens is creating a high barrier. If OpenClaw cannot achieve breakthroughs in algorithm efficiency or large model API pricing reductions, its application scenarios will be limited to high-value, low-frequency fields, unable to penetrate mass long-tail scenarios.

For small and medium enterprises and individual developers, if automation benefits cannot offset token costs, “raising lobsters” becomes a luxury rather than a productivity upgrade. This cost imbalance will further widen the technological gap—only well-funded giants can afford large-scale automation, while ordinary users can only watch from afar.

Therefore, the token issue is not just a technical optimization problem but a critical factor for the business model’s viability. Without effective solutions to token cost-effectiveness, OpenClaw’s widespread adoption will face severe restrictions and may end up being a “hot topic with poor turnout.”