What do you like best about Coder?
It's mostly about choice of flexibility. Unlike locked-in competitors, Coder lets you bring your own tools. The most helpful about Coder is performance and onboarding with:
Native support for Cursor, Windsurf, and all JetBrains IDEs.
Centralized AI Gateway for governed, multi-model LLM access.
UI / UX: Clean, intuitive dashboard that makes complex Terraform deployments feel like a simple app store.
1. pre-built workspaces: "Startup times used to be the 'CDE killer,' but Coder’s Prebuilt Workspaces mean I’m in my code in under 10 seconds. It detects my configuration automatically and has the container 'warm' and ready.
2. Instant onboarding: We recently onboarded five engineers in one morning. Instead of spending two days debugging local environment variables and Docker networking, they just 'claimed' a template and were pushing code before lunch. Easy integrations.
3. seamless networking:
Upsides:
1.Air-Gapped Security
2. Cloud Cost Optimization: The Auto-stop and Resource Quotas features saved us 40% on our cloud bill in the first quarter. We no longer have 'zombie' EC2 instances running over the weekend because someone forgot to turn off their dev box.
3. The Iaas (terraform) approach: Since it's built on terraform, we can version control our entire dev env. Review collected by and hosted on G2.com.
What do you dislike about Coder?
The "Honest Truth": What I Dislike About Coder
The "Terraform Tax" is real
Look, I love the flexibility, but let’s be real: Coder is essentially a full-time job for whoever has to maintain the templates. If you aren't a Terraform wizard, you’re going to struggle. I’ve spent way too many Friday afternoons debugging a broken HCL script because a provider updated and suddenly half the team couldn't spin up their environments. It’s "Infrastructure as Code," which is great until the code breaks and your entire dev team is sitting on their hands.
The "Cold Start" problem
Even in 2026, with all the "Pre-build" hype, the startup times can be a total buzzkill. If I just need to hop in for a quick five-minute hotfix, waiting 90 seconds for a workspace to provision feels like an eternity. Compared to something like Gitpod or a local Docker setup, Coder feels "heavy." It’s a tank—powerful, but it takes a while to get the engine turning.
Maintenance Overhead
Since we self-host for security, we are the product support. When the underlying Kubernetes cluster acts up or a volume gets stuck in a "terminating" loop, it’s on us to fix it. There’s no "support chat" that can reach into our VPC and save us. If you’re a small team without a dedicated Platform or DevOps person, Coder might actually slow you down more than it helps.
The Dashboard feels a bit "Industrial"
The UI is fine, but it’s definitely built by engineers for engineers. It lacks that polished, snappy "SaaS feel" you get with GitHub Codespaces. Sometimes finding a specific workspace setting or viewing logs feels like you’re digging through a file cabinet. It’s functional, but it’s definitely not "pretty."
The AI Gateway setup is a slog
I love that we have an AI Gateway now, but man, setting it up is a manual chore. You have to hand-map every model, set up the rate limits, and configure the fallbacks yourself. I wish there was a "just make it work" button for the AI features instead of having to architect the whole routing logic from scratch.
Summary of the Downsides:
Steep Learning Curve: You need to be a Terraform expert to get the most out of it.
Infrastructure Responsibility: You’re responsible for the uptime of your own dev environment.
Latency: It’s slower to boot than ephemeral, browser-based alternatives.
Enterprise Pricing: The jump from the open-source version to Enterprise is a massive pill to swallow for mid-sized startups.
Bottom line: If you want "easy," go with Codespaces. If you want "total control" and don't mind getting your hands dirty with YAML and HCL for the rest of your life, then Coder is your tool. Just know what you're signing up for. Review collected by and hosted on G2.com.
