<p>Smart contracts are self-executing programs deployed on the blockchain to handle complex business logic. Due to the immutable nature of blockchain, once deployed, smart contracts cannot be modified, making them vulnerable to malicious exploitation if they contain programming flaws. Such vulnerabilities pose significant risks to the security and reliability of blockchain systems. Existing approaches typically rely on fixed, manually defined rules or target single vulnerabilities, limiting scalability as the number and variety of smart contracts and vulnerability types grow. In this paper, we propose a model that addresses these limitations by enabling simultaneous detection of multiple vulnerabilities, acknowledging their potential co-existence within a single contract. The model constructs a control flow graph (CFG) from the bytecode's opcode and assigns edge weights based on control flow sequence features extracted from the source code. This structure enables effective learning of complex control flow paths. By employing multilabel classification, our model is capable of simultaneously detecting six distinct types of vulnerabilities. Experimental results demonstrate that the proposed approach achieves efficient and accurate multi-vulnerability detection in smart contracts.</p>