Dora NodaPull up the GitHub API for coollabsio/coolify and Dokploy/dokploy right now and you get 58,143 stars and 35,500 stars. Every 2026 comparison post repeats some version of the same line: Coolify is the mature incumbent, Dokploy is "the fastest-growing project in the category." One of those two claims is a live number. The other is a tagline that's been copy-pasted across enough listicles that nobody seems to have re-checked it against the actual repos recently. So this is that check — real numbers, pulled today, against what the two projects have actually shipped in the last few months.
The Stars, Checked Today
Both Coolify and Dokploy solve the same problem: turn a bare VPS into a git-push-to-running-app platform, self-hosted, no vendor lock-in. Coolify (coollabsio, created 2021) is the older, broader project — a 280+-service marketplace and multi-server orchestration built up over roughly five years. Dokploy (Dokploy, created April 2024) is younger by three years and has spent that shorter runway betting on a leaner core instead of marketplace breadth.
Multiple comparison articles published around March 2026 cite the same baseline: Coolify at roughly 51,000 stars (then on v4.0.0-beta.468), Dokploy at roughly 31,000 (v0.28.8). Here's that baseline against a live pull from api.github.com/repos/coollabsio/coolify and api.github.com/repos/Dokploy/dokploy, run today:
| March 2026 (reported) | July 8, 2026 (live pull) | Δ absolute | Δ % | |
|---|---|---|---|---|
| Coolify | ~51,000 | 58,143 | +7,143 | +14.0% |
| Dokploy | ~31,000 | 35,500 | +4,500 | +14.5% |
Two caveats worth stating plainly. First, the March figures are secondhand — they're what other sites reported, not a number this post pulled independently, because a precise historical star-timeline (via GitHub's per-star starred_at timestamps) requires an authenticated API token this environment doesn't have. Second, this is one baseline point compared against one live point, not a full weekly time series. It's a spot-check on a specific claim, not a growth curve in the statistical sense.
With that said — the spot-check is still worth running, because the claim it's checking is stated as flatly as any number in this post.
Stars aren't the only community signal, and they don't agree with each other perfectly here either. Coolify reports 575+ contributors and 325,000+ users as of March 2026, with over 20,000 members in its Discord. Dokploy reports over 200 contributors and 6M+ cumulative downloads of its install script. Different metrics, not directly comparable — but the shape is consistent with the star counts: Coolify is the larger community on every axis, Dokploy is meaningfully large for a three-year-old project, and neither number supports a "Dokploy has overtaken Coolify" reading.
Is Dokploy Really Growing Faster?
+14.0% versus +14.5%. That's the entire gap between "mature incumbent" and "fastest-growing project in the category" once you put both projects' own reported numbers next to each other over the same four-month window. It's Dokploy ahead — but by four-tenths of a percentage point, not the landslide the "fastest growth rate" framing implies when it shows up unqualified in a listicle.
Read the other direction, the same numbers say something the growth-rate framing obscures entirely: Coolify's larger base means +14.0% is 7,143 new stars, versus Dokploy's 4,500. In absolute terms — the number of new developers a week who star the repo, poke at the docs, maybe spin up an instance — Coolify is still pulling ahead, not behind. A smaller project can post a higher percentage while a larger one adds more total people, and both of those things are true here at the same time.
None of this erases that Dokploy has real momentum — its April 2026 release cycle (more on that below) was substantial, and a three-year-old project holding within half a point of a five-year-old one's growth rate is a legitimately strong showing. But "fastest growth rate in the category," repeated as settled fact across a stack of 2026 comparison posts, doesn't survive contact with a same-day check of the two repos it's about. Star counts move in bursts around releases and HN front-page moments, not smoothly — so the honest read is "close, and volatile," not "Dokploy is winning."
What the Trajectories Bet On: Marketplace Breadth vs. Resource Footprint
The star-count question is close. What each project actually shipped in the last few months is not — and it's a cleaner signal of where each is placing its bets.
| Coolify | Dokploy | |
|---|---|---|
| Marketplace | 280+ one-click services | 388+ templates |
| Idle RAM | ~1.2GB | ~0.8GB |
| Fresh install | ~2-5 min | ~3 min |
| Latest major release | v4.0.0, April 2026 (redesigned UI, multi-server orchestration) | v0.29.0, April 18, 2026 (AI debugging, MCP + CLI, shared git providers) |
Coolify's pitch has always been breadth: if you need Supabase, Plausible, or n8n pre-templated, it's probably already there, and the 280-service count reflects a five-year head start on template contributions. Dokploy's pitch is that most teams don't need 280 templates — they need a fast install, a small footprint, and a small number of primitives that don't get in the way. The RAM gap (0.8GB vs. 1.2GB idle) is small in absolute terms but matters on the $5-10/month VPS tier both projects are actually built for, where 400MB is the difference between comfortably running a Postgres instance alongside your app and swapping.
There's a third data point that cuts against both of those stereotypes: MCP server capability. Coolify's official instance-level MCP server (merged May 2026, PR #9862) ships exactly 10 tools — list/get calls for infrastructure, servers, projects, applications, databases, and services. It's read-only; write support is listed as "planned." Dokploy's MCP server, part of the same v0.29.0 release that shipped its CLI, exposes 508 tools across 49 categories with full read/write coverage, alongside a 449-command CLI — meaning an AI agent talking to Dokploy today can actually create, deploy, and manage infrastructure, not just inspect it. The five-year-old marketplace leader currently has the thinner AI-agent integration. That's not what "mature incumbent vs. leaner newcomer" predicts, and it's the sharpest concrete evidence in either direction about what each team is actually prioritizing right now.
The Ceiling Both Hit
Whatever the star count or the MCP tool count says this week, both projects share an architectural fact that neither has solved: there's one control-plane instance, and it's a single point of failure.
Coolify's own docs describe multi-server support as one server running the dashboard, the internal Postgres database, Git webhook handling, and deployment orchestration to every other connected server — if that instance goes down, so does your ability to deploy or manage anything, even though your already-running apps keep running. An open GitHub issue on the project (#7384) also documents that the multi-server UI currently blocks setup entirely once a volume mount is involved, on top of Docker Swarm clustering being labeled experimental. Dokploy's architecture, per its own docs, follows the identical shape: a central instance manages remote deploy and build servers over SSH, with v0.29.0 adding non-root multi-server support — a real improvement, but to the same single-control-plane model, not a different one.
Neither project frames this as a bug so much as a boundary: both are built, correctly, as single-VPS-first tools, and the SSH-orchestrated control plane is a reasonable design for that scope. But it means neither has an answer for "what happens when the machine running Coolify (or Dokploy) itself dies" beyond "restore from backup and reconnect your servers." That's the layer a fleet-lifecycle platform built on Cluster API is designed to remove entirely — because the control plane is itself a Kubernetes cluster, HA is the default, not an unsolved feature request.
Bex.co is the open-source, AI-native Render alternative built on that model — push a git repo, get a running HTTPS service on machines you own, with a Cluster API control plane instead of a single SSH-orchestrated daemon. Star the repo on GitHub or deploy your first app today.
What the Numbers Actually Say
Strip away the taglines and the two trajectories say something more interesting than either "Coolify wins" or "Dokploy is disrupting it." The star-count race is close enough that "fastest growth in the category" is marketing copy more than measured fact — Coolify is still adding more total developers per week off a larger base, even as Dokploy holds a slightly better percentage. Feature velocity doesn't track project age either: the newer, leaner project currently ships the more capable AI-agent integration, while the older one leads on template breadth and ecosystem depth built up over years. Docker-host teams choosing between them today aren't picking a clear winner — they're picking which bet (marketplace coverage or a small, fast, increasingly agent-friendly core) fits their VPS budget and their tolerance for a single point of failure neither project has removed yet.
Sources:
- GitHub — coollabsio/coolify (live API pull, July 8, 2026)
- GitHub — Dokploy/dokploy (live API pull, July 8, 2026)
- Coolify vs Dokploy comparison, nextgrowth.ai
- Coolify vs Dokploy: Complete Comparison Guide 2026, Contabo
- Dokploy v0.29.0: AI-Powered Debugging, MCP Server & CLI, Shared Git Providers
- Coolify MCP Server documentation
- Coolify GitHub Issue #7384 — multi-server volume mount restriction
- Multiple Servers — Coolify Docs
- Dokploy Multi-Server Management — DeepWiki