Deploy LiteLLM on CometVPS
One OpenAI-format proxy in front of 100+ LLM providers. Issue per-user virtual keys with budgets, track spend per request, and fail over between providers — all self-hosted on your own server.
What is LiteLLM?
LiteLLM is a self-hosted proxy that puts a unified OpenAI-compatible API in front of every major LLM provider. Apps talk to one endpoint; LiteLLM handles the provider-specific quirks, key management, fallbacks, rate limits, and per-user cost tracking.
One API, Every Model
Virtual Keys & Budgets
Visibility & Control
Key Features
Everything you need to run a production AI gateway on your own infrastructure
100+ LLM Providers
Unified OpenAI-format API across OpenAI, Anthropic, Google, Azure, Bedrock, Groq, OpenRouter, Ollama, vLLM, and 100+ others. One SDK, every model.
Virtual Keys & Budgets
Issue per-user or per-team virtual API keys with rate limits, monthly spend caps, and allowed-model lists. Revoke instantly without rotating provider keys.
Cost Tracking & Analytics
Track tokens, cost, and latency per request, per user, per model. Postgres-backed dashboards make AI spend finally visible across your org.
Fallbacks & Load Balancing
Round-robin, weighted, or latency-based routing across model deployments. Automatic fallback to a backup model when a provider rate-limits or fails.
Installation Guide
Get LiteLLM running on your CometVPS server in 5 simple steps
Security Tip
Always set a strong LITELLM_MASTER_KEY environment variable and never hand it out — it's the root key. Issue per-user virtual keys for everything else. Front the proxy with HTTPS, and store provider API keys in environment variables (not in config.yaml committed to git). Rotate keys regularly.
Recommended VPS Plans
LiteLLM is light on resources — pick a plan based on traffic volume
Core VPS - Basic
2 vCPU Cores, 4GB RAM, 60GB NVMe
LiteLLM is lightweight — a small VPS happily proxies thousands of requests per day for personal use and side projects.
Supernova VPS - Spark
4 AMD Ryzen Cores, 8GB DDR5, 100GB NVMe
Faster Ryzen cores and DDR5 keep latency low under team load, with room for the Postgres database holding usage logs and virtual keys.
Supernova VPS - Flare
6 AMD Ryzen Cores, 16GB DDR5, 200GB NVMe
Plenty of capacity for a company-wide AI gateway routing across many providers, with detailed cost analytics in Postgres.
Why Deploy on CometVPS?
Get the best performance and reliability for your AI gateway
Centralized API Keys
NVMe SSD Storage
10Gbps Network
24/7 Expert Support
Ready to Deploy LiteLLM?
Centralize your AI keys, track spend per user, and route around provider outages — all from one self-hosted proxy.