Are amicu and Ollama competitors?

Only partly. Ollama is a local runner. amicu is a deploy tool for llama.cpp and vLLM servers, locally or on GPU providers. They can be used side by side.

Does amicu use Ollama?

No. amicu starts llama.cpp or vLLM itself. It does not depend on Ollama.

Can I run models on my own machine with amicu?

Yes. On macOS, Linux and Windows, amicu runs llama.cpp or vLLM as a user service on this machine, with a loopback endpoint.

amicu vs Ollama

Ollama is an easy way to download and run open models on your own computer, with its own API and model library. amicu is a deploy tool: you describe models, hardware and deployments in HCL, review a plan, and amicu runs llama.cpp or vLLM on your machine or on GPU providers such as Modal and RunPod, tracking what it started. If you want to chat with a model on your laptop, Ollama is simpler. If you want a repeatable, reviewable deployment, possibly on a remote GPU, amicu fits. They can be used side by side.

Quickstart All comparisons

Side by side

amicu vs Ollama
	Ollama	amicu
What it is	A way to run open models locally or in the cloud, with a model library.	A CLI that deploys llama.cpp or vLLM servers from HCL configuration.
Getting started	Install it and pull a model.	Install the CLI, run init, then validate, plan and apply.
Engine	Its own runtime; its README lists llama.cpp as a supported backend.	llama.cpp or vLLM, chosen per deployment.
Where models run	Your own computer, or its cloud models offering.	This machine, or GPU cloud providers such as Modal and RunPod.
Configuration	Commands and Modelfiles.	Declarative HCL kept in your repository, with a plan before every change.
API	Its own REST API, with OpenAI-compatible and Anthropic-compatible clients.	The runtime's own OpenAI-compatible server; amicu is not a gateway.
Auth on remote endpoints	Local by default; securing a network-exposed server is up to you.	Off-machine endpoints require an api_key secret, and apply verifies anonymous requests are rejected.
Tracking what is running	The Ollama server manages loaded models.	State records every started instance so destroy can stop it.
License	MIT.	AGPL-3.0-only.

Sources for the Ollama column: Ollama documentation, Ollama on GitHub. Details can change; check the official documentation.

When to choose Ollama

You want to try open models quickly on your own computer.
You use desktop apps or tools that integrate with Ollama directly.
You do not need vLLM or a reviewable deployment workflow.

When to choose amicu

You want deployments defined in HCL, reviewed in a plan, and tracked in state.
You need vLLM, or a remote GPU on a provider such as Modal or RunPod.
You want endpoint authentication and residency rules checked as part of apply.

Questions

Are amicu and Ollama competitors?: Only partly. Ollama is a local runner. amicu is a deploy tool for llama.cpp and vLLM servers, locally or on GPU providers. They can be used side by side.
Does amicu use Ollama?: No. amicu starts llama.cpp or vLLM itself. It does not depend on Ollama.
Can I run models on my own machine with amicu?: Yes. On macOS, Linux and Windows, amicu runs llama.cpp or vLLM as a user service on this machine, with a loopback endpoint.

Try amicu

Install the CLI and deploy a model on your machine or a GPU provider.

Quickstart Providers