Abstract Hybrid c c g of quantum numbers 1−− are considered in a quark model with constituent quarks and gluon. The lowest JP=1− states may be built in two ways, lg=1 (gluon excited) corresponding to an angular momentum between the gluons and the c c system, while l c c =1 (quark excited) corresponds to an angular momentum between the c and the c. The lowest lying hybrid JP=1− state in the flux tube model is similar to the lg=1 in the quark-gluon model. In particular it verifies the selection rule that it cannot decay into two fundamental mesons. The l q q =1 hybrid may decay into two fundamental mesons, but with decay widths larger than 1 GeV, which tells that they do not really exist as resonant states. using a chromoharmonic potential, we find no mixing between the lg=1 and l c c =1 . More realistic potentials might induce a strong mixing between them, implying that no hybrid meson exist. If, on the contrary, such a strong mixing does not occur, we find, in agreement with the flux tube model, that only the lg=1 appears as a real resonant state. In such a case, hybrid mesons may exist as resonances only if they are decoupled from the ground state channel, which explains the difficulty to observe them experimentally. We reconsider accordingly the Ono–Close–Page scenario of mixing between charmonium and charmed-hybrid to explain the anomalies around 4.1 GeV. We find a very small mixing between radially excited charmonium and hybrid mesons, which forbids considering the ψ(4.040) and ψ(4.160) as combinations of 3S charmonium and lg=1 hybrid meson with a large mixing.