Hardened Ollama, LiteLLM & Self Hosted LLM Server

AWS Application Stacks
Base
Hardened build
minimal ports, security patches applied at build time
Access
Unique credentials
generated on first boot, readable only by root
Verified
Boots working
services pass a health gate before release
Support
24/7, 365 days
by email and live chat, 24 hour response SLA

Overview

cpp Server, LocalAI, AnythingLLM and 5 more, CPU-only and patched, with per-instance admin passwords and API keys, HTTPS, and no access to instance metadata.

Why the cloudimg image

This image arrives installed, configured and hardened, so there is no manual setup before you can use it. It is built on a patched base, runs with credentials generated uniquely for your instance on first boot, and passes an automated health check before every release. Every deployment is backed by 24/7 support from cloudimg engineers.

Key features

  • No default keys: admin passwords, API keys and secrets are generated uniquely on each instance's first boot, and every web UI or API is served over HTTPS with a per-instance certificate
  • Model processes cannot reach the instance metadata service, so a prompt or a configured URL can never fetch the instance's AWS credentials; internal services bind to localhost
  • Ollama, LiteLLM, llama.cpp Server, LocalAI, AnythingLLM, Bifrost, Infinity and more, CPU-only: pick one as the software version at launch

Description

This is a repackaged open source software product wherein additional charges apply for security hardening, secure-by-default configuration and per-instance setup.

What cloudimg adds to every version

  • Every image is built by cloudimg on a fully patched base with the latest security updates applied at build time.
  • Each product is CPU-only (no GPU drivers or CUDA libraries ship), generates its administrator password, API keys and secrets uniquely on the first boot of each instance, serves its web interface or API over HTTPS with a per-instance certificate, and its services cannot reach the instance metadata service, so a prompt or a configured URL can never fetch the instance's credentials.

Included products

  • AnythingLLM 1.14.1: chat with your documents and local or remote models
  • Bifrost 1.6.6: a fast OpenAI-compatible LLM gateway
  • big-AGI 2.0.5: a multi-model chat interface
  • GPT-Load 1.4.9: an LLM API gateway with key pooling
  • Infinity 0.0.77: an embeddings and reranking API
  • LiteLLM 1.88.0: an OpenAI-compatible gateway for 100+ LLM APIs
  • llama.cpp Server b9937: runs GGUF models on CPU behind an OpenAI-compatible API
  • LocalAI 4.4.3: an OpenAI-compatible API for local models
  • NextChat 2.16.1: a lightweight chat interface
  • Ollama + AnythingLLM 1.18.0
  • Ollama 0.31.2: runs open models on CPU behind a simple API

How to choose a version

  • Run open models on CPU behind an API: Ollama, llama.cpp Server or LocalAI
  • One OpenAI-compatible gateway in front of many LLM APIs: LiteLLM, Bifrost or GPT-Load
  • A chat interface for your team: AnythingLLM, NextChat or big-AGI
  • An embeddings and reranking API: Infinity

Network and access defaults

  • Internal services are bound to localhost.

This listing is a collection of separate AMIs, one per product and version, chosen from the Software version dropdown at launch.

Support

cloudimg provides 24/7 technical support for every version in this listing by email and live chat, 365 days a year, with a 24-hour response SLA: support@cloudimg.co.uk.

Trademarks

Ollama, LiteLLM and the other product names are trademarks of their respective owners. cloudimg is not affiliated with or endorsed by any of them.

Related technologies

ai gatewayopenai compatible apillama.cpplocalaianythingllmbifrostgpt-loadinfinity embeddingsnextchatbig-agiprivate aichat uicpu inferencemodel servinggenerative ai