A few days ago, I used OpenClaw to distill the thinking patterns of several well-known figures and enabled a feature that allowed responses to be generated from their perspectives.
The result was immediate.
My token consumption skyrocketed, and the MiniMax subscription I was using quickly reached its limit.
That left me with two choices.
The first was to upgrade to a larger MiniMax plan.
The second was to try a different large language model.
My friend ZJ had already told me that MiniMax was the best value-for-money model he had ever used.
To be honest, I was skeptical.
So I consulted ChatGPT, asked several other AI systems, read reviews, and compared options. Eventually, I decided to purchase 100 yuan worth of DeepSeek API credits, a model that many people described as having outstanding cost performance.
I told ZJ about my decision.
He replied that he had used DeepSeek before and didn't think it was as capable as MiniMax.
Once again, I remained unconvinced.
Perhaps he had used an older version, I thought. Maybe the latest version had improved significantly.
After all, technology evolves quickly.
But after spending a day with DeepSeek, I found myself gradually agreeing with him.
First, there was the cost.
At my current level of usage, I was spending around 10 yuan per day. That meant my 100-yuan balance would only last about ten days.
It was far less economical than I had expected.
Then there was the experience itself.
Its coding ability was respectable.
However, when it came to writing, I was deeply disappointed.
The structure, depth, and overall quality of its articles fell well below my expectations. In many cases, it struggled to produce a complete blog post that I would feel comfortable publishing.
The difference between it and MiniMax was obvious.
Do I regret making that purchase?
Not at all.
Because there are many things in life that cannot truly be understood until you experience them yourself.
ZJ is one of my closest friends.
I know his advice comes from sincerity and good intentions.
Yet I still refuse to accept anyone's opinion as an absolute answer.
Not because I distrust them.
But because no one is infallible.
Other people's experiences can serve as valuable references.
They cannot replace your own.
In fact, this reminds me of ZJ himself.
His website domain is antisubmissivist.com.
He once told me that the word "antisubmissivist" was his own creation, meaning someone who resists blind conformity.
The more I think about it, the more I realize that we are probably the same kind of person.
We listen to advice.
But we do not blindly obey it.
We respect experience.
But we still want to verify things for ourselves.
Even when others turn out to be right, we want to arrive at that conclusion through our own journey rather than simply borrowing theirs.
This thought led me to another question: parenting.
How much should parents interfere in their children's lives?
Should they guide them away from every mistake?
Or should they allow them to stumble, fail, and learn through experience?
Of course, parents should do their best to protect children from mistakes that could permanently damage their future.
But for smaller failures and detours, I believe experience is often the better teacher.
A single mistake can leave a deeper impression than a hundred warnings.
A single setback can teach more than an entire shelf of self-help books.
Life does not come with a perfect answer key.
Even wrong decisions can have value.
Because the process of exploration, trial, and error is itself a form of growth.
Others may show you the direction.
But the true nature of the road can only be understood once you walk it yourself.