I'm sorry, but this question makes no sense at all. Define what you mean by "reliable" in this context. Are you asking if it will ALWAYS produce the correct answer to any problem posed in code? If that is your question, then you need to define what the correct answer would be. NO tool is perfectly reliable, if used by someone who does not understand the tool. In this respect, that means you need to understand the mathematics/theory behind what you are doing when you write code. In many cases, that means an understanding of the numerical methods.
If, by reliable you mean is it consistent, always producing the same result for different people, given the "same" code? Again, that is not really a valid question, since some code will be based on random numbers. You may get subtly different results basd on a random seed. And in other cases, you may get VERY different results. But the word "reliability" is meaningless in that context.
If by reliable, do you mean it will always produce a mathematically correct result? Well, no. It is trivially easy to pose a problem that has no valid result obtainable by mathematics, or by any computing language. Again, this means you need to understand the mathematics when you use a tool like MATLAB. You need to understand the theory of what you are doing. Would you put a child in the driver's seat of your automobile, and expect a good result to arise? As well, it is easy to pose a problem where computer code will fail, due to numerical issues. As I said, you may need an understanding of numerical methods to use such a tool.
Do users need to modify MATLAB, according to their needs? Again, that is a question with no clear answer. Yes, EVERYONE (who uses the language for any period of itme) tends to write functions that do what they tend to need to do. That makes MATLAB work better, for their personal use. But in my eyes, that is not a question of reliability. It is merely the way that anyone tends to use any such programming environment.
Anyway, if I look for the definition of reliable, I see this:
consistently good in quality or performance; able to be trusted.
"a reliable source of information"
Google, referencing Oxford Languages dictionaries as its source.
Can MATLAB be trusted in that sense? Yes, and no. If you write complete crap for code, then you should expect complete crap for results. What can I say? A tool is no better than its user. Is MATLAB trustable, in the sense that the code has been written by trusted professionals in the art, and has been tested to every extent possible? Well, yes. Of course, no code is provably perfect. So asking if MATLAB is 100% perfectly trustable, that is surely asking too much.
