Jaringan Tor merupakan jaringan yang memberikan anonimitas bagi penggunanya, namun sering kali dikritik karena performanya yang lebih rendah, terutama dalam hal latensi dan kecepatan unduh dibandingkan jaringan non-Tor. Dalam penelitian ini, file dengan tiga ukuran berbeda (50KiB, 1MiB, dan 5MiB) diunduh melalui kedua jaringan, dan waktu unduh tersebut dicatat. Data tersebut dianalisis menggunakan parameter statistik kuartil pertama (Q1), median (MD), dan kuartil ketiga (Q3) untuk mengidentifikasi perbedaan performa di antara kedua jenis jaringan. Penelitian menggunakan data Tor Metrics periode 11 April 2017 hingga 30 Januari 2025 dan mengungkap bahwa jaringan Tor memiliki tolerance factor 2-3,6 kali lebih lambat dibandingkan jaringan publik. Overhead tertinggi mencapai 310% pada file 5MiB-Q1, dengan standar deviasi latensi Tor mencapai 34ms dibanding 22ms pada jaringan publik. Analisis menunjukkan fixed overhead berupa setup relay, enkripsi berlapis, dan routing tiga hop berdampak signifikan pada file kecil namun relatif kurang dominan pada file besar. Variabilitas performa Tor yang tinggi disebabkan oleh inkonsistensi kondisi jaringan dan karakteristik inheren arsitektur multi-hop. Hal ini menunjukkan bahwa latensi dan arsitektur multi-hop Tor berdampak langsung terhadap efisiensi proses unduh, sehingga perlu dipertimbangkan dalam penggunaan aplikasi yang sensitif terhadap waktu. / Tor is widely recognized for its user anonymity features but is often criticized for lower performance, especially in terms of latency and download speed. In this research, files of three different sizes (50KiB, 1MiB, and 5MiB) were downloaded through both networks, and the download times were recorded at various intervals. The data was analyzed using the first quartile (Q1), median (MD), and third quartile (Q3) statistical parameters to identify performance differences. This study utilizes Tor Metrics data from April 11, 2017, to January 30, 2025, and reveals that the Tor network has a tolerance factor of 2-3.6 times slower compared to public networks. The highest overhead reaches 310% for 5MiB-Q1 files, with Tor latency standard deviation reaching 34ms compared to 22ms on public networks. Analysis shows that fixed overhead consisting of relay setup, layered encryption, and three-hop routing significantly impacts small files but is relatively less dominant for large files. High Tor performance variability is caused by network condition inconsistencies and inherent characteristics of multi-hop architecture. This demonstrates that latency and multi-hop architecture of Tor directly impact download process efficiency, requiring consideration in applications sensitive to time.