RF Coaxial Cable Quality and Frequency Rating

There is an EM limit to coax, which is where waveguide modes manifest -- that is, signals no longer propagate in the TEM00 (current lengthwise on core, opposite on shield, radial electric field, tangential magnetic field -- the standard coax field diagram). Which is to say, the distance between core and shield, or the shield diameter or circumference, is comparable to 1/4 wavelength or more (dimensions and wavelength fraction depending on mode). This still isn't a hard cutoff, signals still pass -- but they don't pass in the orderly fashion that we expect from coax in normal (below-cutoff) use, that is, the impedance and velocity factor are all over the place; plus mode conversion occurs (due to asymmetries in the cable construction or orientation, energy couples into orthogonal modes, where it gets absorbed, causing transmission notches, impedance peaks/valleys, etc.). Which is further "a thing" because not all modes are coupled by the core-shield connection.

More often, there is a practical limit to coax, which is the attenuation rate per length, which increases dramatically with frequency, particularly for the poorer quality / simply constructed kinds like RG-58.

Generally, larger coax has lower losses, right up to (or near, anyway) mode breakup. Cell phone and television transmitters for example use inch-scale conduit, for signals from say a few hundred MHz, to low GHz; the dielectric might be mostly air, with a loose plastic support structure to hold the core in place, maximizing core diameter and thus conductive surface. When your electrical bills correlate directly with transmission line losses, investing in such specialized, low-loss materials is desirable.

Or compare RG-58 to RG-174, basically the same stuff, scaled differently; the latter has significantly more attenuation/length.

By the way, there are proper and ready sources for all of these data. Understand that Wikipedia is at best an aggregator of knowledge. YouTube videos and random web articles are user created so depend entirely on who wrote them, from what sources, and why (and, sources are almost never given, so you have no way of checking how well they've represented it). Popular videos are rarely well sourced/cited, and well sourced/cited videos are rarely popular (and guess which one YT is likely to return from searching?). Volunteer systems, like Wikipedia, Stackexchange, etc., have varying standards (e.g. Wikipedia requires references, if not necessarily good ones; only some Stacks do); and generally "good enough" passes, subject to social norms (e.g. it's not worth correcting something if it's going to "rock the boat"). Which when it comes to complex topics that are opinioned far more often than they are understood, can be quite frustrating to see; take basically any EMC topic on here, for example.

Even worse, there's AI large language models, which are designed to mimic written language, with (currently) little to no basis of knowledge, nor capacity for inference upon that knowledge (it's no accident that GPT, etc. sourced answers are currently prohibited here). Needless to say, don't attempt to source information from such services. Use them for what they're good at: language itself, not the things language represents.

Whatever the matter; in this case (cables and connectors), standards exist, which manufacturers follow and customers expect. Simply go shopping for cables and connectors, and look up datasheets! Datasheets are primary sources; units are given; and you can find definitions if you need to.

You may need supporting info to understand some things (like, what the heck are the various impedances within a multiconductor cable?), or a deeper understanding to intuit them (aha, the impedances are between wires in different configurations, e.g. parallel (common mode) or inverse (differential) pairings; and the impedance, inductance and capacitance are related by fundamental constants, material properties and geometry). These are good directions for further research. A number of these theoretical topics are explored in depth in university courses, which you may also find helpful. (Not to try and read too deeply into your level of knowledge, of course.)

Of course, manufacturers themselves span a range of quality. They might simply lie on the datasheet, or not provide one at all -- or the supplier you're buying them through doesn't provide such. Buying things on generic marketplaces (like Amazon, eBay, Ali, etc.) is very much caveat emptor. The temptation to sort by price is a strong one, but a risk laid barest through these sources. There are industry-recognized channels to go through; to source a particular brand-name manufacturer, find their list of authorized distributors. Parts quality, like knowledge itself, is a traceable path of connections, relationships; and like knowledge itself, quality is an ongoing process, a verb: knowledge is not something you learn once, but understand that memory itself decays over time, the assumptions that underlaid it change, and it must be forever checked and re-checked for truth.