LiteLLM: one address for every model

This template does not use a GPU: you can also run it on a dedicated CPU machine if you would rather have more cores and memory than a graphics card.

Video memory

não precisa de GPU

Setup

~2 min

Access

port 4000

Billing

hourly, in BRL

LiteLLM puts a single API in front of everything: your own model, vendor A's model and vendor B's. The application calls one address, and you swap the model behind it without touching code.

LiteLLM Proxy unifies all LLM providers behind one OpenAI-compatible API: OpenAI, Anthropic, Bedrock, Gemini, Azure, local vLLM/TGI. Cost tracking, rate-limit, virtual keys, central logs. Essential for enterprises.

What it is for

• Change models without changing the application
• Spend caps and per-team or per-customer keys
• Central log of every call, for auditing
• Automatic failover when one model goes down

How to deploy

1. Create your account and add balance (card or Pix, no subscription).
2. In the console, pick the LiteLLM Proxy template and a machine — the console hides the ones that do not meet the requirement.
3. In about 2 minutes the setup finishes and the access address shows up in the panel, on port 4000.

Done? Just destroy the machine and billing stops with it. No contract, no minimum commitment.

FAQ

Does it need a GPU?

No. LiteLLM only routes; pick a CPU machine and point it wherever the model runs.

Does it work with a local model?

Yes, that is the most common setup: it sits in front of a vLLM or Ollama on the same network.

Can I cap spend per user?

Yes, with virtual keys that have their own limits and expiry.

Related templates