On GitHub, the neondeex/atlas-filemanager project was published — the Atlas file manager. The author calls it "a file manager at the speed of thought": native, with GPU rendering, and files never leave your machine. The repository has 121 commits, the code is open, and builds target Windows 10/11. The status is a free alpha, with the "rough edges" and fast iterations mentioned right in the README.
Inside there are five display modes, and the view is chosen separately for each panel: Grid, List, Details, Miller columns, Tree. Tabs and splits in any direction, dragging files between panels, restoring a closed tab. Command palette — Ctrl+Shift+P, jump to folder with fuzzy search — Ctrl+K, preview via Space for images, video, audio, documents, code, and archives. ZIP, RAR, 7z, and TAR open as folders, without extraction. Conversion of images, video, audio, and documents happens locally, from the context menu or a radial picker. Clouds (Google Drive, OneDrive, and others) are connected via rclone, alongside SMB, FTP, and WebDAV. There are password-protected private folders, tags, profiles, and disk cleanup. Those who want to build it themselves can clone the repository and run cargo run --release; you'll need MSVC build tools with the Windows SDK and Rust. The toolchain version is pinned in rust-toolchain.toml, and the first build takes a long time pulling GPUI.
Windows Explorer has lived in a strange status for years: everyone has it, and almost no one is happy with it. The alternative market — Total Commander, Directory Opus, Files — rests on habit and plugins. Atlas comes at it from a different angle: not "more features," but "faster rendering." Our assessment: the bet on GPU and Rust here isn't marketing, but a concrete attempt to fix large directories, where the built-in Explorer noticeably stalls.
What does this change in practice? If the alpha really maintains smoothness on folders with tens of thousands of files, some people will consider switching file managers for the first time in years. For your wallet, entry is free, Premium is a one-time payment, not a subscription — a pleasant rarity. But the limits are hidden in panels, profiles, cloud connections, and batch renaming, so active users will have to pay up.
Who wins? Those for whom the built-in Explorer isn't enough, and everyone who keeps files locally and doesn't want to hand them over to the cloud. Who loses — we believe, first of all, the heavyweights like Directory Opus: their strength lies in years of accumulated fine details, but not in interface speed. True, they have support and polish behind them, while Atlas has an alpha and an open repository.
The source has no version, no build number, no release date — only a README and a file tree. There are no benchmarks: the claims about "instantly opening folders" are backed by nothing, and without numbers on large directories it's just a promise. The Premium price, the size of the limits, and how mature the alpha is are unknown. The README on GitHub is the project's own voice, so speed and stability will have to be tested on your own machine.
Should you replace the built-in Windows Explorer with an alpha from a small team — or are habit and predictability more important than speed?
A one-time Premium license for expanded limits: an honest model for an everyday tool, or a hidden fee for basic things like extra panels?
GPU rendering in a file manager — a real pain point that needs solving, or engineering beauty the user won't notice?
Prediction: if within a quarter Atlas doesn't show public benchmarks on large directories and doesn't gain a noticeable user base, the project will remain a niche experiment for those who love Rust and GPUI.
Source: GitHub / neondeex
Source: github.com · post in Telegram