MVSEP is a web-based service that leverages artificial intelligence technology to decompose audio files into separate elements, such as vocals and musical instruments. Users are able to upload audio files with a maximum size of 100MB in various formats and choose from a selection of AI models for the separation process. The platform provides both free and premium subscription options, where premium users gain access to enhanced models that can isolate additional components, including bass, drums, piano, guitar, and more. The output files are available for download in several formats, including MP3, WAV, FLAC, and M4A. This service is especially advantageous for musicians, producers, and audio engineers who aim to extract specific audio elements for purposes such as remixing, analysis, or practice. Currently, MVSEP operates with two distinct datasets; the first is a synthetic dataset designed to evaluate separation quality between vocals and instruments, while the second is a diverse multisong dataset featuring individual tracks across various genres, capable of identifying vocals, instrumentals, bass, drums, and additional elements. Moreover, the platform continuously seeks to enhance its capabilities and expand its offerings to better serve its user community.