Bot hosting guide
How to Host a Telegram Bot on a VPS
Telegram bots are light. The hard part is keeping them online, deployable, and secure without overpaying for hardware.
Affiliate disclosure: some links below are affiliate links. We may earn a commission at no extra cost to you.
What a Telegram bot actually needs
Most bots are small programs that receive updates, call an API, and reply. They spend most of their time waiting, not computing. A single-core VPS with 512 MB to 1 GB RAM can comfortably run several lightweight bots.
Memory grows when you add a database, image processing, browser automation, or a local AI model. Start small and watch memory usage for a week before deciding to upgrade.
Webhook vs long polling
Long polling asks Telegram for new updates in a loop. It is simple, needs no domain, and works behind NAT. The downside is a constant request loop and slightly higher latency.
Webhook lets Telegram push updates to your HTTPS endpoint. It is more efficient and scales better, but requires a public domain, a TLS certificate, and a reverse proxy. For production bots, webhook is usually the better long-term choice.
If you are just testing, start with polling. Move to webhook when you add a domain or expect steady traffic.
How much server do you need?
| Setup | RAM | vCPU | Notes |
|---|---|---|---|
| 1 simple bot | 256–512 MB | 1 | Polling, no database |
| 2–5 bots | 1 GB | 1 | Shared process manager or Docker |
| Bot + database | 1–2 GB | 1–2 | Postgres or SQLite depending on load |
| Bot + browser automation | 2–4 GB | 2 | Playwright and Chromium are memory hungry |
| Bot + local LLM | 8 GB+ | 4+ | See the GPU guide for inference |
Deployment checklist
- Create a non-root user and disable password SSH login.
- Store the bot token and API keys in environment variables, not in the repository.
- Run the bot under systemd or Docker with a restart policy.
- Enable a firewall and expose only the ports you need.
- For webhook mode, put Nginx or Caddy in front and use TLS.
- Log to a file or the system journal, and add basic error alerts.
- Set up automatic security updates.
A systemd unit keeps the bot alive after logout and restarts it after a crash. Docker achieves the same with a restart policy and makes dependencies reproducible.
Uptime and monitoring
Telegram bots fail silently. Add a simple health endpoint or a periodic self-check that messages you when something breaks. Uptime monitors that ping an HTTP endpoint are enough for webhook bots; for polling bots, log a heartbeat and alert if it stops.
Keep an eye on memory. A slow leak will eventually crash the process, and the restart policy will hide the problem until it happens often enough to notice.
Security basics
- Never commit tokens or API keys. Rotate them if exposed.
- Validate and rate-limit user input to prevent abuse.
- Run the bot with the least privileges it needs.
- Keep the OS and dependencies patched.
- Back up any database before upgrades.
What it costs
A single bot fits on the cheapest VPS from most providers, roughly $4–6 per month. Hetzner offers the best value if an EU region is acceptable. Vultr and DigitalOcean are convenient when you need a specific region or a familiar ecosystem.
DigitalOcean
Best documentation if you are new to servers. $6 gets a basic droplet.
FAQ
How much RAM does a Telegram bot need?
A simple bot runs in 128–512 MB of RAM. A 512 MB or 1 GB VPS is enough for one to three lightweight bots. Add memory if you use a database, browser automation, or machine learning.
Should I use webhook or long polling?
Webhook is more efficient and scales better because Telegram pushes updates to your server. Long polling is simpler to set up and works without a public HTTPS endpoint. For production, webhook with a reverse proxy is usually the better choice.
Do I need a domain and SSL for a bot?
Only for webhook mode. Telegram requires a public HTTPS endpoint for webhooks. A free TLS certificate from Let's Encrypt and a small reverse proxy are enough. Long polling needs neither.
How do I keep the bot running after I disconnect?
Run the bot as a systemd service or inside Docker with a restart policy. Both keep the process alive after logout and restart it automatically after crashes or reboots.
Last updated: November 2026. Always check current provider pricing and your local regulations before deploying.