The answer is complicated. Any reading of Kant’s massive corpus of writings initially gives one a mass of conflicting liberal and anti-liberal claims.
He did argue for a robust freedom of speech, at least for the philosophical use of reason. He advocated property rights. He resisted some types of commodifications of human beings. And on numerous matters in foreign policy he called for more peace-oriented and humanistic policies.
Yet he also argued against allowing the smallpox vaccination: “So that states do not become overcrowded with people and thus stifled from the outset, two evils are placed in them as antidotes: smallpox and war.” More trivially, he was against allowing individuals to sell their hair to wigmakers (although buying wigs was allowable; Kant wore wigs). More seriously, he believed it was not a crime for an unmarried mother to kill her child, since bastards are illegitimately brought into the polity. He complained that “a prolonged peace favours the predominance of a mere commercial spirit, and with it a debasing self-interest, cowardice, and effeminacy, and tends to degrade the character of the nation” (note the “mere commercial spirit” barb, which has implications for Kantian economic policy). He held that the important teachings in educational institutions should be subject to the sanction of the state. He was ambivalent about the separation of church and state and allowed for the censorship of some church teachings. And he denied strenuously the right of individuals to rebel against evil rulers, no matter how tyrannical.
Also, given Kant’s many negative remarks about Jews, women, and blacks and other races, it is far from clear that he believes the majority of human beings are capable of agency at a level worthy of moral dignity. Of course, those negative remarks can be bracketed in Kant’s system as mere anthropology and discounted to lower status than his philosophical pronouncements. But the remarks do undercut his connection to the classically liberal ideal of universal freedoms or rights.
Still, we have a puzzle. Many subsequent liberal thinkers have been energized by Kant’s formulations of principles of respect for human dignity and treating individuals as ends in themselves. But they have also been shocked by his frequent anti-liberal principles and policies. Were the anti-liberal deviations merely the result of his being a man of his times, as he was born and raised in a strict Pietist Lutheran family in 1700s Prussia, after all? Was he simply inconsistent, since he wrote on thousands of topics across many decades? Is there some complicated philosopher’s way of making his whole corpus more or less consistent that deep scholarship can reveal?
I will argue that the anti-liberalism is much deeper in Kant’s philosophy than the liberalism. That means saying something about the ringlingly liberal-sounding principles that are indeed integral to Kant’s philosophy. That something is this: One must always interpret a comprehensive philosopher’s remarks on applied matters in the context of his philosophical system.
Let me give an exaggerated, fictional example to stress this point about method. Suppose one reads Chapter One of a philosopher’s book in which one finds affirmations that individuals should seek freedom and realize their true selves and that only in liberty can one find dignity and so on. In Chapter Two, one finds the philosopher arguing that one’s body is not one’s true self; rather one has an immaterial soul that is one’s real Self and that it comes to full actualization only upon separating from this physical realm. Chapter Three argues that when one’s true self is actualized, it does so by merging one’s self into an uber-Self of souls that are collectively real, and that one’s highest moral obligation is to achieve such a merger. As one works through Chapter Four, one learns that the philosopher believes that back in this material world the official state/church institutions are the temporal embodiment of one’s collectively true Self. Finally, in Chapter Five one reads the philosopher’s perfectly logical conclusion that individual freedom means obeying the state/church’s orders.
The example is fictional, but it makes the point about interpretation: What some philosophers mean by self, individual, and freedom is the opposite of what other philosophers mean. So before settling upon an interpretation of a philosopher’s normative claims one always must investigate his or her more fundamental claims in metaphysics and epistemology. Many of us begin our study of Kant by reading his Groundwork as undergraduates, and we may interpret his formulations of the Categorical Imperative as being continuous with the liberal individualist tradition coming out of John Locke, Voltaire, and Adam Smith. But when we pursue Kant’s big books—most importantly the three Critiques—we find ourselves in an entirely different philosophical universe.
And especially in the case of Kant, incorporating his fundamental distinction between phenomenal and noumenal realms is essential.
Kant argues that we experience only the phenomenal world with its features of time, space, cause and effect. This world is bounded, finite, and Isaac-Newton-physically deterministic. Yet beyond the phenomenal world is the noumenal, which is perhaps the realm of God, freedom, and immortality.
Kant also believes that if there is to be morality, it must come from a place of freedom. Consequently, that freedom must be outside of the phenomenal world, since the phenomenal is ruled by deterministic cause and effect in space and time. In other words, since morality depends upon freedom and the phenomenal realm is unfree, the origin of morality can only be noumenal.
The human self must then be a microcosm of the phenomenal/noumenal split. We are of course phenomenal beings, subject to natural law. But to be moral beings, an aspect of us must be free. We must have a noumenal self that is not subject to natural laws.
So when Kant says that human selves are moral agents, that they have a capacity for dignity, and that their freedoms ought to be respected, he is speaking only about noumenal selves. He is not speaking about our phenomenal selves. Our phenomenal selves are not free and are consequently not in the realm of morality.
Further: Within Kant scholarship there is controversy over whether Kant meant for us to believe that noumenal selves really exist in a noumenal realm or whether he is merely positing them as regulative ideas. A regulative idea, in Kant’s philosophy, is one that we cannot experience or verify empirically but which our reason must posit if it is to make sense of some important philosophical project such as knowledge and morality. The regulative-idea interpretation has traction, since we learn from Kant’s first Critique that any knowledge of the noumenal is totally closed off to us. So we must accept that a noumenal self may or may not actually exist—it may be merely a regulative idea that we posit philosophically in order to make sense of morality, if morality actually is possible.
In either case, freedom is a defining feature of the noumenal self, and bondage is a defining feature of the phenomenal self. And if only my noumenal self is free, and freedom is required for moral agency and responsibility, then my phenomenal self is not a morally responsible agent.
But it is precisely one’s phenomenal self that is one’s lived self—the self that makes a living through physical labor and commerce, that has sex and bears children, that consumes food and wears clothes, that listens to music and works out. That self, in the Kantian ontology, has no freedom at all. So it has no moral agency. And, Kant argues, it is therefore properly subject to a supposed metaphysical nugget—i.e., a noumenal self—in another dimension that issues commands to it.
Where does this leave the Kantian case for liberalism? To conclude this line of argument bluntly:
If our case for political liberty ultimately depends on following orders from a hypothetical voice from another realm, then we don’t have much of a case.
My recommendation: If we are looking for a principled philosophical justification of liberalism, then bypassing Kant is a good idea.
 Metaphysics of Morals, Doctrine of Right 2, General Note E1, Akad. 6, 335f.
 Critique of Judgment, Section 28.
 Metaphysics of Morals, “General Remarks on the Legal Consequences of the Nature of the Civil Union,” Political Writings, 2nd edition, edited by Hans Reiss (Cambridge University Press, 1991), pp. 143-145.
 E.g., “the euthanasia of Judaism is the pure moral religion.” Kant, Streit der Fakultaten, in Werke, 11:321, cited in Paul Lawrence Rose, Revolutionary Antisemitism from Kant to Wagner (Princeton University Press, 1990), p. 96.
 “It is easy to analyse man; but woman betrays her secrets even though she is unable to keep those of others (owing to her love of gossip). Man is fond of domestic peace and submits easily to its governance so as to be unmolested in his business. Woman has no dislike for domestic war for which she is armed with her tongue … .” Kant, Anthropology from the Pragmatic Point of View (1798), quoted in Léon Poliakov, The Aryan Myth: A History of Racist and Nationalist Ideas in Europe (Meridien, 1977), p. 171.
 Kant opposed racial inter-breeding: “The mingling of stocks (due to great conquests), little by little erodes the character and it is not good for the human race in spite of any so-called philanthropy,” quoted in Poliakov, p. 172. See also Wulf D. Hund, “‘It must come from Europe’ The Racisms of Immanuel Kant,” 2011, p. 91. https://www.academia.edu/5234906/_It_must_come_from_Europe_._The_Racisms_of_Immanuel_Kant.
 Or perhaps not, as a variant upon it is one standard reading of Hegel’s philosophy.
 See this formulation in The Stanford Encyclopedia of Philosophy entry on Kant: “My noumenal self is an uncaused cause outside of time, which therefore is not subject to the deterministic laws of nature in accordance with which our understanding constructs experience.” Rohlf, Michael, “Immanuel Kant,” The Stanford Encyclopedia of Philosophy (Spring 2016 edition), Edward N. Zalta (ed.), URL = http://plato.stanford.edu/archives/spr2016/entries/kant/, Section 5.2.
 Regulative ideas: “concepts created by the faculty of reason that cannot be met in experience but that can serve as models or goals for actual human behavior.” Frederick Rauscher, “Kant’s Social and Political Philosophy,” The Stanford Encyclopedia of Philosophy (Fall 2016 edition), Edward N. Zalta (ed.), URL = <http://plato.stanford.edu/archives/fall2016/entries/kant-social-politic…;.