With the rapid development of 5th Generation Mobile Communication Technology (5G), the increased bandwidth not only bolsters communication efficiency but also paves the way for high-precision location-based services within these industrial applications. Positioning by using a single base station (BS) possesses potential vital values in scenarios that multiple BSs are not applicable, such as unmanned aerial vehicle assisted emergency communication scenarios, dynamic and temporary industrial 5G scenarios, etc. However, due to a lack of observed measurements from different base stations, the high accurate and stable positioning by using a single BS faces challenges. To address this problem, this paper introduces a novel approach utilizing the open-source 5G platform OpenAirInterface (OAI) to design application programming interfaces (APIs) for accessing low layer data. By extracting data directly from actual protocol stacks, we have developed a multidimensional data fusion localization system based on Vision Transformer (ViT) in a single-BS mode named FLIT. Conducted within an office setting, our experiments spanned a 100MHz bandwidth, culminating in a system capable of achieving an average positioning error of just 6.9cm, with 90% of measurements within a 9cm error margin and 80% within 5.3cm, thus attaining centimeter-level accuracy in positioning.