The aMule project officially released version 3.0.0 in June 2026, marking the first major update to the open-source P2P client in over five years. The previous release, version 2.3.3, had been sitting untouched since February 7, 2021, leaving users on Linux and macOS with a client that had fallen well behind eMule on Windows in terms of raw performance.
The team labeled this new version “The alive again version,” and the name is not an exaggeration. This is not a minor patch or a routine maintenance update, it is a structural overhaul that rewrites critical performance components, replaces the entire build system, modernizes the dependency stack, and cleans up years of accumulated legacy code.
The headline figure the developers are leading with is striking: peer-to-peer download speeds on the same hardware are between 100 and 380 times faster than aMule 2.3.3 across macOS, Linux, and Windows.
On macOS running Apple Silicon, benchmarks went from 0.35 MB/s to 135 MB/s, an improvement of 381 times. On Linux ARM, speeds jumped from 0.34 MB/s to 117 MB/s, roughly 345 times faster. These numbers were measured over a 90-second window on a local network, with a single peer downloading a 30 GB file from a Linux seeder running the same version under test.
The technical fix behind the speed leap
The biggest single contributor to these gains is the port of eMule’s CUploadDiskIOThread system into aMule. Previously, disk reads and eD2k packet construction were handled on the main thread, creating a bottleneck every time the client needed to read from disk or assemble packets for transfer.
Moving that work off the main thread is what produced the dominant share of the upload performance gains, according to the official changelog. Additional leecher-side fixes, including partfile-write offload and throttler corrections, stacked another 4 to 6 times improvement on top of that.
The result is that aMule 3.0.0 now sustains approximately 4.8 times the upload throughput of eMule 0.70b on Windows, reaching 106 MB/s compared to eMule’s 22 MB/s in the same test environment. On the download side, aMule 3.0.0 reached 39 MB/s versus eMule’s 20 MB/s, roughly 1.9 times faster.
This matters because back in 2021, users were already reporting that aMule 2.3.3 was dramatically slower than eMule on the same links and settings, cases where eMule exceeded 4.5 MB/s while aMule could not break 1 MB/s. Version 3.0.0 directly addresses that long-standing gap.
Two additional bugs with real user impact were also corrected. Setting MaxUpload to 0, which is supposed to mean unlimited, was actually capping upload speed at the current rate plus 5 KB/s. MaxDownload was not functioning as a literal speed cap either, it operated as a ratio controller rather than enforcing an actual limit. Both throttlers have been completely rewritten.
The release also includes improved Kad searches with parallel queries, fresh native packages for every major platform, Windows installer and portable version, Universal DMG for macOS, AppImage and Flatpak for Linux, and full protocol compatibility with existing eMule-based clients is preserved throughout.
What this means for the P2P sharing community
To understand why this release matters beyond the benchmark numbers, it helps to remember what eMule and aMule actually represent in the history of file sharing. These clients did not operate on BitTorrent.
They ran on the eD2k and Kademlia networks, and the reason those networks survived while Napster and eDonkey2000 were shut down is that they evolved from a server-based architecture, vulnerable to legal intervention, to the Kademlia network, a fully decentralized peer-to-peer system with no central server to target. That architectural shift made it effectively impossible to shut down, and it kept the network alive through two decades of industry pressure.
The eD2k and Kad ecosystem maintained a particularly strong user base in Europe through the 2000s, where eMule and aMule became deeply embedded in internet culture. Beyond file sharing, these networks were early proof that distributed traffic without centralized infrastructure was viable, an idea that resurfaced later in BitTorrent and various distributed storage systems.
The network never truly died, but the software maintaining access to it had stagnated. aMule 2.3.3 was five years old, and eMule’s official stable release had also gone without meaningful updates for a long time, leaving an active user base with aging tools.
What aMule 3.0.0 represents for that community is a client that finally matches the performance expectations of modern hardware. The bottlenecks that had made aMule noticeably slower than eMule for years have been resolved. The build system has been modernized. Native binaries are available for current platforms and architectures, including ARM64.
The project, which had gone quiet long enough that many considered it abandoned, is now back under active development. Whether this brings a meaningful wave of returning users to the eD2k and Kad networks is a separate question, streaming services have replaced many of the use cases that originally drove people to P2P, and BitTorrent has a commanding lead in the file sharing space. But for the community that never left, and for anyone curious enough to return, aMule 3.0.0 is the most capable version of this client that has ever been released.
Are you an eMule or aMule user? Do you think this comeback can bring classic P2P back to life? Let us know in the comments!