A year ago we wrote Winget vs Chocolatey vs Scoop and recommended starting with winget. Twelve months and a lot of installs later — did that hold up? Here's the receipts.
TL;DR
| Metric | 2025 | 2026 | Change |
|---|---|---|---|
| Winget catalog | ~7,500 | ~10,300 | +38% |
| Chocolatey community | ~10,000 | ~10,400 | +4% |
| Avg time from publisher release to winget manifest | 18 hrs | 6 hrs | -67% |
Open issues in microsoft/winget-pkgs |
2.1k | 1.4k | -33% |
| winget feature requests merged | 89 | 124 | +39% |
Winget is shipping faster than Chocolatey can match. The recommendation still stands: start with winget.
What got better in winget this year
Catalog grew 38%
The winget catalog finally crossed Chocolatey on raw package count in early 2026. More importantly: the long tail improved. Niche tools that used to require Chocolatey now have winget manifests — JetBrains everything, Hashicorp suite, almost all CNCF projects, every major language runtime.
Manifest publishing speed
A year ago, a fresh VS Code release took 18 hours on average to land in winget. Today: 6 hours. Most popular apps (Chrome, Firefox, Discord) get same-day manifests because publishers themselves push them.
Install success rate
Microsoft hasn't published official numbers but our user analytics on 50k+ installs in 2025-2026:
| winget | Chocolatey | |
|---|---|---|
| First-try install success | 96.4% | 93.8% |
| Retries needed | 2.1% | 4.0% |
| Hard failures | 1.5% | 2.2% |
The gap comes from winget's hash verification and Microsoft re-hosting the manifest index — fewer "broken upstream URL" issues.
Configuration & quality of life
New in 2026:
winget configure(Configuration File support) — declaratively define machine state with YAML, install + configure togetherwinget upgrade --include-unknownworks reliably for non-winget installs- Better PowerShell module integration
- Native ARM64 manifests for most popular apps
What's still better in Chocolatey
Internal hosting and enterprise
Chocolatey for Business has features winget doesn't match yet:
- Self-hosted private repositories
- Package internalisation (mirror everything to your network)
- Package builder GUI
- Integration with SCCM, Ansible, Puppet
- Audit trails and compliance reporting
If you're a Windows admin managing 500 machines, Chocolatey is still the answer.
PowerShell-first culture
Chocolatey packages are PowerShell scripts. You can:
- Read the install logic before running
- Customise it (chocolateyInstall.ps1)
- Build your own with familiar PS patterns
winget manifests are YAML + opaque InstallerSwitches. Less hackable, but also less footgun.
packages.config in source control
<packages>
<package id="vscode" />
<package id="git" version="2.50.0" />
</packages>Many shops have years of this in Git. Migrating to winget's JSON schema is non-trivial when you have 200 packages × 50 machines.
Better error messages
Chocolatey errors tend to be more descriptive. "Hash mismatch: expected XYZ, got ABC, URL https://..." vs winget's terser "0x80070005: Access denied".
Where winget closed the gap
Things that were Chocolatey wins last year but aren't anymore:
| Feature | Status in 2026 |
|---|---|
| Version pinning | ✅ winget pin is mature |
| Custom install args | ✅ via winget settings |
| Manifest source control | ✅ winget export --include-versions |
| GUI client | ✅ Multiple third-party GUIs (WingetUI, UniGetUI) |
| Configuration-as-code | ✅ winget configure (DSC v3 backend) |
What didn't change
Both tools still:
- Need admin for most installs (no app store sandboxing)
- Pull installers from publisher URLs (subject to publisher uptime)
- Don't manage Windows itself or drivers
- Don't replace package managers for languages (npm, pip, cargo)
And both still depend on the underlying installer (.msi/.exe/.appx) running silently — neither can fix a broken installer upstream.
Specific updated recommendations
Pick winget if you're…
- A solo developer or home user → start here
- Setting up a new machine —
winget importis faster than choco install -y - Building install scripts for non-technical people (the .json schema is portable)
- Anyone whose org doesn't already use Chocolatey
Pick Chocolatey if you're…
- Managing 50+ Windows machines at work
- Already invested in PowerShell-based configuration management
- Need offline / private package hosting
- On a compliance-heavy team (audit logs, signed packages)
Use both if you're…
- A developer at a Chocolatey-mandated workplace — use winget for personal CLI tools, Chocolatey for what IT requires
- Maintaining legacy systems alongside greenfield projects
What about Scoop?
Scoop holds its niche perfectly: portable dev tools, user-mode installs, no admin. We still recommend it as a complement to winget, not a replacement. Use winget for big GUI apps; use Scoop for CLI tools like jq, fd, ripgrep, hyperfine that don't need admin or system integration.
Predictions for 2027
- Winget catalog will hit 15k packages as publisher push increases
- Microsoft will ship a first-party winget GUI (the third-party ones are decent but indicate a gap)
- Chocolatey will pivot harder to enterprise — community catalog growth has plateaued
winget configurewill become mainstream for IaC-style Windows setups
The bottom line
A year ago the answer was "start with winget." Today the answer is "unless you have a specific reason not to, use winget."
Chocolatey isn't going away — it's evolved into the enterprise Windows package manager, which is a perfectly good place to be. But for the rest of us, winget won 2026. We'll check back in 2027.
What's next?
- Winget vs Chocolatey vs Scoop (original 2025 post) → — for context
- Complete Windows 11 setup with winget → — practical setup guide
- Winget commands cheatsheet → — every command
- Best winget packages for developers → — curated list