The abstract of V. Lafforgue's paper https://arxiv.org/abs/1404.6416 says

For any reductive group G over a global function field, we use the cohomology of G-shtukas with multiple modifications and the geometric Satake equivalence to prove the global Langlands correspondence for G in the direction "from automorphic to Galois".

Theorem 1.1 of this paper explains precisely what is meant by this in the split case, and section 2.1 does the same in the non-split case. In short, the space of cusp forms admits a decomposition indexed by global Langlands parameters, compatible with Satake at unramified primes.

In addition to the reverse direction (i.e. the Arthur multiplicity formula), Lafforgue also mentions some further work that might be considered part of the Langlands correspondence:

I think the compatibility with the local Langlands correspondence is still work in preparation of Genestier and Lafforgue.

I think the compatibility with the theta correspondence is still unproved.