We propose a way of computing 4-manifold invariants, old and new, as chiral correlation functions in half-twisted 2d N = (0, 2) theories that arise from compactification of fivebranes. Such formulation gives a new interpretation of some known statements about Seiberg-Witten invariants, such as the basic class condition, and gives a prediction for structural properties of the multi-monopole invariants and their non-abelian generalizations. 5 Non-abelian generalizations 51 5.1 Coulomb-branch index and new 4-manifold invariants 54 A Curvature of the canonical connection on the universal bundle 57 A.1 SU (N f ) invariance and the moment map 59 B Topological twist of SQED 60 1.1 Unorthodox invariants of smooth 4-manifolds Searching for new invariants of smooth 4-manifolds, that potentially could go beyond the Seiberg-Witten and Donaldson invariants, it was proposed in [1] to consider a twodimensional quantum field theory T [M 4 , G] as a rather unusual invariant of smooth structures on a 4-manifold M 4 . Specifically, T [M 4, G] is a 2d N = (0, 2) superconformal theory that, apart from M 4 , also depends on a choice of a root system G and is invariant under the Kirby moves. Luckily, conformal field theories in two dimensions exhibit rich mathematical structure which, on the one hand, is rich enough to (potentially) describe the wild world of smooth 4-manifolds and, on the other hand, is rigorous enough to hope for a precise mathematical definition of the invariant T [M 4 , G]. In fact, for many practical purposes and applications in this paper, a mathematically inclined reader can think of T [M 4 , G] as a functor that assigns a vertex operator algebra (VOA) to a smooth 4-manifold M 4 (and a "gauge" group G).Composing it with other functors that assign various quantities to 2d conformal theories, one can obtain more conventional invariants of smooth 4-manifolds:Here, Z can be any invariant of a 2d conformal theory with N = (0, 2) supersymmetry, e.g., its elliptic genus, chiral ring, moduli space of marginal couplings, or central charge. Since 2d theory T [M 4 ; G] is systematically determined by M 4 and invariant under the Kirby moves, all such invariants lead to various 4-manifold invariants; some are simple and some are quite powerful. In particular, it was conjectured in [1] that chiral ring of the theory T [M 4; G] for G = SU (2) or, equivalently, its Q + -cohomology knows about Donaldson invariants of M 4 . One of the main goals in this paper is to present some evidence to this conjecture and to build a bridge between VOA G [M 4 ] and more traditional 4-manifold invariants. Conjecturally, at least for manifolds with b + 2 > 1, one can trade SU (2) Donaldson invariants for Seiberg-Witten invariants defined in a simpler gauge theory, with gauge group G = U (1), which would be too trivial if not for an extra ingredient, the additional spinor fields. We wish to study how these invariants are realized in 2d theory T [M 4 ; G] with G = SU (2) and G = U (1), respectively. In particular, we shall see that in 2d realizati...