X repriced and restricted its API to a degree that pushed a large amount of research and monitoring work toward browser-based collection. That shift is the reason demand for Twitter proxies and X proxy access exists at all: the supported route became unaffordable for many use cases that previously used it.
At the same time, anonymous browsing of the site has been progressively restricted. How much is visible without a session has changed repeatedly, so any guidance stating a fixed rule about what loads logged out will be wrong within months. Verify the current behaviour rather than trusting a claim of either kind.
The practical consequence is that a proxy solves the network half of this problem and none of the rest. If your workflow needs an authenticated session, that is governed by X's terms, and the account risk is real regardless of how good the exit is.
Be realistic about what this does not fix. If X's terms do not permit your collection, a proxy does not change that, and the platform has enforced against automated access. If the supported API covers your use case at a price you can pay, use it: it is more stable than any browser-based route and carries none of the same risk. Proxies are the pragmatic answer where the API's scope or pricing genuinely does not fit, and where the data is public.
If you're comparing workflows before you commit, explore all proxy use cases to see which setup fits best.