Jet flavour tagging is crucial in experimental high-energy physics. A tagging algorithm, - , is presented, which exploits a transformer-based neural network that is substantially faster to train than state-of-the-art graph neural networks. The algorithm uses information from particle flow-style objects and secondary vertex reconstruction for - and -jet identification, supplemented by additional information that is not always included in tagging algorithms at the LHC, such as reconstructed $$K_{S}^{0}$$ and $$\Lambda ^{0}$$ and $$K^{\pm }/\pi ^{\pm }$$ discrimination. The model is trained as a multiclassifier to identify all quark flavours separately and performs excellently in identifying - and -jets. An -tagging efficiency of $$40\%$$ can be achieved with a $$10\%$$ -jet background efficiency. The performance improvement achieved by including $$K_{S}^{0}$$ and $$\Lambda ^{0}$$ reconstruction and $$K^{\pm }/\pi ^{\pm }$$ discrimination is presented. The algorithm is applied on exclusive $$Z \rightarrow q\bar{q}$$ samples to examine the physics potential and is shown to isolate $$Z \rightarrow s\bar{s}$$ events. Assuming all non- $$Z \rightarrow q\bar{q}$$ backgrounds can be efficiently rejected, a $$5\sigma $$ discovery significance for $$Z \rightarrow s\bar{s}$$ can be achieved with an integrated luminosity of $$60~\text {nb}^{-1}$$ of $$e^{+}e^{-}$$ collisions at $$\sqrt{s}=91.2~\textrm{GeV}$$ , corresponding to less than a second of the FCC-ee run plan at the boson resonance.