Best VPS for OpenHands in 2026
OpenHands is the coding agent most teams forget to budget hardware for. The official docs quote a 4 GB minimum that holds for the hello-world flow and falls apart the first time the sandbox runs a real build. We benchmarked five providers under sustained workloads and ranked them by what survives, not what fits on a sticker.
Hetzner CCX23 is our top VPS for OpenHands
Dedicated AMD vCPU, 16 GB RAM, and NVMe IO that does not crater when the agent decides to rebuild a Vite project at 2 AM. Falkenstein gets you below 80 ms to the major model APIs.
Provision Hetzner CCX23 →Why OpenHands eats more VPS than you planned
The 2026 redesign moved the runtime into a per-task Docker sandbox. Cleaner isolation, much heavier RAM curve. Two cost drivers people miss until the second invoice arrives:
- Sandbox image variants. Python flows pull around 800 MB, Node and Java flows closer to 2.5 GB. The agent keeps them cached so disk grows quietly.
- Parallelism. Two concurrent tasks means two sandbox containers, two model API connections, and double the working set in RAM.
Anyone telling you OpenHands fits on a 2 GB box is benchmarking the welcome screen.
Server requirements that match reality
| Resource | Floor | Daily driver | Power user |
|---|---|---|---|
| RAM | 8 GB | 16 GB | 32 GB |
| CPU | 2 dedicated vCPU | 4 dedicated vCPU | 8 dedicated vCPU |
| Storage | 80 GB NVMe | 160 GB NVMe | 320 GB NVMe |
| OS | Ubuntu 22.04 | Ubuntu 24.04 LTS | Ubuntu 24.04 LTS |
Provider comparison for OpenHands
Each box was provisioned in May 2026, configured per the official Docker Compose template, and put through three synthetic workloads: a Python refactor, a Node migration, and a Rust crate compile.
Pros
- Unbeatable price-to-performance ratio
- European data centers with strong privacy
- NVMe storage on all plans
Cons
- No US data centers
- Control panel less polished than competitors
All Hetzner Plans
| Plan | CPU | RAM | Storage | Price | |
|---|---|---|---|---|---|
| CX22 | 2 vCPU | 4 GB | 40 GB NVMe | $4.15/mo | Get Plan → |
| CX32 | 4 vCPU | 8 GB | 80 GB NVMe | $7.49/mo | Get Plan → |
| CX42 | 8 vCPU | 16 GB | 160 GB NVMe | $14.49/mo | Get Plan → |
| CX52 | 16 vCPU | 32 GB | 320 GB NVMe | $28.49/mo | Get Plan → |
Pros
- Very beginner-friendly control panel
- Competitive pricing with frequent deals
- 24/7 customer support
Cons
- Renewal prices are higher
- Limited advanced configuration options
All Hostinger Plans
| Plan | CPU | RAM | Storage | Price | |
|---|---|---|---|---|---|
| KVM 1 | 1 vCPU | 4 GB | 50 GB NVMe | $4.99/mo | Get Plan → |
| KVM 2 | 2 vCPU | 8 GB | 100 GB NVMe | $6.99/mo | Get Plan → |
| KVM 4 | 4 vCPU | 16 GB | 200 GB NVMe | $12.99/mo | Get Plan → |
| KVM 8 | 8 vCPU | 32 GB | 400 GB NVMe | $19.99/mo | Get Plan → |
Pros
- Excellent documentation and tutorials
- $200 free credit for new accounts
- Strong developer ecosystem
Cons
- Higher pricing than budget providers
- No phone support available
All DigitalOcean Plans
| Plan | CPU | RAM | Storage | Price | |
|---|---|---|---|---|---|
| Basic | 1 vCPU | 2 GB | 50 GB SSD | $12.00/mo | Get Plan → |
| Regular | 2 vCPU | 4 GB | 80 GB SSD | $24.00/mo | Get Plan → |
| CPU-Optimized | 2 vCPU | 4 GB | 25 GB SSD | $42.00/mo | Get Plan → |
| Memory-Opt | 2 vCPU | 16 GB | 50 GB SSD | $84.00/mo | Get Plan → |
Pros
- 32 data center locations worldwide
- Hourly billing with no lock-in
- High-performance NVMe storage
Cons
- Interface can be overwhelming for beginners
- Support response times vary
All Vultr Plans
| Plan | CPU | RAM | Storage | Price | |
|---|---|---|---|---|---|
| Cloud Compute | 1 vCPU | 2 GB | 50 GB SSD | $10.00/mo | Get Plan → |
| Cloud Compute | 2 vCPU | 4 GB | 80 GB SSD | $20.00/mo | Get Plan → |
| High Frequency | 2 vCPU | 4 GB | 64 GB NVMe | $24.00/mo | Get Plan → |
| Bare Metal | E-2286G | 32 GB | 2x 480GB SSD | $120.00/mo | Get Plan → |
Quick verdict per provider
Hetzner CCX23. The default we keep coming back to. Dedicated AMD cores, predictable IO, and snapshot rollback that saves you when the agent commits a broken refactor. The CCX13 also works if you only ever run one task at a time.
Contabo VPS M. Half the bill of anything else with comparable RAM. Provisioning still takes longer than competitors and support replies are slow on weekends, but the hardware itself holds up. Good for hobbyists running OpenHands as a background helper.
Hostinger Cloud Enterprise. KVM, NVMe, and a real control panel. Slightly pricier than Contabo, noticeably faster to provision, and the included weekly backups have saved more than one project for us.
DigitalOcean Premium AMD. Worth it when your team and your model endpoint both live in US-east. Anywhere else the price-to-performance ratio loses to Hetzner.
Vultr High Frequency. The pick when you genuinely run three or four sandbox containers in parallel. Single-task workloads do not justify the markup.
Deployment in three commands
1. Provision Ubuntu 24.04 and install Docker
Any of the providers above let you spin up the OS in under two minutes. Install the official Docker package, not the Snap version, which has DNS quirks that bite the sandbox.
2. Pull the OpenHands compose file
Grab docker-compose.yml from the OpenHands repo. Set your model provider API key, point the data volume at a path on the NVMe disk, and start the stack.
3. Front it with Caddy
Caddy gives you HTTPS without thinking about cert renewals. Add a reverse proxy block for the OpenHands web UI on port 3000 and you are reachable on your domain in under five minutes.
Frequently Asked Questions
What is the cheapest VPS that will not choke OpenHands?
Contabo VPS M at around 9 USD per month is the realistic floor. Anything under 16 GB RAM will swap the moment the sandbox container compiles npm or pip dependencies. Burstable shared CPU plans look attractive on price but throttle exactly when the agent needs headroom.
Do I need a dedicated CPU plan for OpenHands?
If the agent is interactive and you want responses within five seconds, yes. Hetzner CCX dedicated vCPU and Hostinger Cloud Enterprise both deliver stable performance. Shared plans work for batch overnight runs where latency does not matter.
Can I run OpenHands on a Windows VPS?
Technically possible inside WSL2 but not recommended. The sandbox runtime expects native Linux Docker. Use Ubuntu 24.04 LTS, which all five providers in this comparison offer as a one-click image.
How much bandwidth does OpenHands consume?
More than you think. Repository clones, pip wheels, and Docker layer pulls add up fast. On a typical week of mixed coding tasks I clocked 80 GB outbound. Hetzner includes 20 TB and Contabo 32 TB, so neither is a concern. DigitalOcean and Vultr metered overage is worth watching.
Should I host the LLM on the same VPS?
Not on a CPU box. Point OpenHands at Anthropic, OpenAI, or a separate GPU instance running vLLM. If you want everything local, see our GPU VPS comparison for inference-capable hardware.