ramalama
Synopsis
ramalama [options] command
Description
The goal of RamaLama is to make AI boring.
The RamaLama tool facilitates local management and serving of AI Models.
On first run, RamaLama inspects your system for GPU support, falling back to CPU support if no GPUs are present.
RamaLama uses container engines like Podman or Docker to pull the appropriate OCI image with all of the software necessary to run an AI Model for your systems setup.
Running in containers eliminates the need for users to configure the host system for AI. After initialization, RamaLama runs AI Models within containers based on OCI images. RamaLama pulls a container image specific to the GPUs discovered on the host system. These images are tied to the minor version of RamaLama. For example, RamaLama version 1.2.3 on an NVIDIA system pulls quay.io/ramalama/cuda:1.2. To override the default image, use the --image option.
RamaLama pulls AI Models from model registries, then starts a chatbot or a REST API service with a single command. Models are treated similarly to how Podman and Docker treat container images.
When both Podman and Docker are installed, RamaLama defaults to Podman. The RAMALAMA_CONTAINER_ENGINE=docker environment variable can override this behavior. When neither is installed, RamaLama attempts to run the model with software on the local system.
On macOS systems that use Podman for containers, configure the Podman machine to use the libkrun machine provider. The libkrun provider enables containers within the Podman machine to access the Mac GPU. See ramalama-macos(7) for further information.
On systems with NVIDIA GPUs, see ramalama-cuda(7) to correctly configure the host system.
RamaLama CLI defaults can be modified via ramalama.conf files. Default settings for flags are defined in ramalama.conf(5).
FEDORA SILVERBLUE AND TOOLBOX
On Fedora Silverblue and other immutable variants the system is read-only. You can run RamaLama in either of these ways:
-
Toolbox (recommended): Create a Toolbox container and install RamaLama inside it (e.g.
pip install ramalamaordnf install ramalama). Whenflatpak-spawnis available (included in Fedora Toolbox by default), RamaLama automatically detects the toolbox environment and usesflatpak-spawn --hostto run container engine commands (Podman/Docker) on the host. GPU/accelerator detection tooling (such asnvidia-smi,nvidia-ctk,npu-smiandmthreads-gmi) is also run on the host, and host CDI configuration (e.g./etc/cdi) is read via the toolbox's/run/hostmount, so GPU acceleration is detected and used transparently even when the driver tooling is only installed on the host. No manual socket bind-mounting or engine configuration is needed. TheToolboxfield inramalama infooutput indicates whether toolbox mode is active. -
Host Installation: Install RamaLama on the host via
rpm-ostree install ramalamaif the package is available for your image.
The model store defaults to ~/.local/share/ramalama, which is writable on Silverblue. Temporary files are stored under ~/.cache/ramalama/tmp in toolbox environments (instead of /var/tmp) to ensure they are accessible to container engine commands running on the host.
SECURITY
Test and run your models more securely
Because RamaLama defaults to running AI models inside rootless containers using Podman or Docker, these containers isolate AI models from information on the underlying host. With RamaLama containers, the AI model is mounted as a volume into the container in read-only mode. This keeps the runtime process (llama.cpp or vLLM) isolated from the host. Since ramalama run uses the --network=none option, the container cannot reach the network and leak information out of the system. Finally, containers are run with the --rm option, which means content written during execution is removed when the application exits. Hosted API transports such as openai:// bypass the container runtime entirely and connect directly to the remote provider; those transports inherit the provider's network access and security guarantees instead of RamaLama's container sandbox. For that reason the hosted API transports are deprecated and will be removed in a future release; use ramalama chat --url <endpoint> --api-key <key> --model <model> to reach a remote provider instead.
Here's how RamaLama delivers a robust security footprint:
- Container isolation: AI models run within isolated containers, preventing direct access to the host system.
- Read-only volume mounts: The AI model is mounted in read-only mode, so processes inside the container cannot modify host files.
- No network access:
ramalama runuses--network=none, so the model has no outbound connectivity through the container. - Auto-cleanup: Containers run with
--rm, wiping temporary data once the session ends. - Dropped Linux capabilities: No Linux capabilities are granted to attack the host.
- No new privileges: Linux kernel controls prevent container processes from gaining additional privileges.
MODEL TRANSPORTS
RamaLama supports multiple AI model registries types called transports. Supported transports:
| Transports | Prefix | Website |
|---|---|---|
| URL based | https://, http://, file:// | https://web.site/ai.model, file://tmp/ai.model |
| HuggingFace | huggingface://, hf://, hf.co/ | huggingface.co |
| ModelScope | modelscope://, ms:// | modelscope.cn |
| Ollama | ollama:// | ollama.com |
| rlcr | rlcr:// | ramalama.com |
| OCI Container Registries | oci:// | opencontainers.org |
Examples: quay.io, Docker Hub, Artifactory |
Models specified in the Hugging Face <org>/<model> format are automatically pulled from Hugging Face. Models can also be specified using a shortname (e.g. tiny) which is resolved via shortnames.conf, or with an explicit transport prefix such as huggingface://, oci://, ollama://, https://, http://, or file://.
The default transport can be overridden in the ramalama.conf file or via the RAMALAMA_TRANSPORT environment variable. Running export RAMALAMA_TRANSPORT=huggingface changes RamaLama to use the HuggingFace transport for unqualified model names.
URL support means if a model is on a web site or even on your local system, you can run it directly.
ramalama pull huggingface://afrideva/Tiny-Vicuna-1B-GGUF/tiny-vicuna-1b.q2_k.gguf
ramalama run file://$HOME/granite-7b-lab-Q4_K_M.gguf
To make it easier for users, RamaLama uses shortname files, which contain aliases for fully specified AI Models. RamaLama reads shortnames.conf files if they exist. These files contain name/value pairs for model definitions. The following table specifies the order in which RamaLama reads these files. Any duplicate names override previously defined shortnames.
| Shortnames type | Path |
|---|---|
| Distribution | /usr/share/ramalama/shortnames.conf |
| Local install | /usr/local/share/ramalama/shortnames.conf |
| Administrators | /etc/ramalama/shortnames.conf |
| Users | $HOME/.config/ramalama/shortnames.conf |
$ cat /usr/share/ramalama/shortnames.conf
[shortnames]
"tiny" = "hf://TheBloke/TinyLlama-1.1B-Chat-v1.0-GGUF"
"granite" = "hf://ibm-granite/granite-3.3-8b-instruct-GGUF"
"granite:7b" = "hf://instructlab/granite-7b-lab-GGUF"
"ibm/granite" = "hf://ibm-granite/granite-3.3-8b-instruct-GGUF"
"merlinite" = "hf://instructlab/merlinite-7b-lab-GGUF"
"merlinite:7b" = "hf://instructlab/merlinite-7b-lab-GGUF"
...
GLOBAL OPTIONS
--debug
Print debug messages.
--dryrun
Show container runtime command without executing it (default: False).
--engine
Run RamaLama using the specified container engine. Default is Podman if installed, otherwise Docker.
The default can be overridden in the ramalama.conf file or via the RAMALAMA_CONTAINER_ENGINE environment variable.
--help, -h
Show this help message and exit.
--nocontainer
Do not run RamaLama workloads in containers (default: False).
The default can be overridden in the ramalama.conf file.
OCI images cannot be used with the --nocontainer option. This option disables automatic GPU acceleration, containerized environment isolation, and dynamic resource allocation.
--quiet
Decrease output verbosity.
--runtime=llama.cpp | vllm
Specify the runtime to use. Valid options are llama.cpp and vllm (default: llama.cpp).
The default can be overridden in the ramalama.conf file.
--store=STORE
Store AI Models in the specified directory (default rootless: $HOME/.local/share/ramalama, default rootful: /var/lib/ramalama).
The default can be overridden in the ramalama.conf file.
--version, -v
Show the program version and exit.
COMMANDS
| Command | Description |
|---|---|
| ramalama-bench(1) | benchmark specified AI Model |
| ramalama-benchmarks(1) | view and interact with historical benchmark results |
| ramalama-chat(1) | OpenAI chat with the specified REST API URL |
| ramalama-containers(1) | list all RamaLama containers |
| ramalama-convert(1) | convert AI Models from local storage to OCI Image |
| ramalama-daemon(1) | run a RamaLama REST server |
| ramalama-info(1) | display RamaLama configuration information |
| ramalama-inspect(1) | inspect the specified AI Model |
| ramalama-list(1) | list all downloaded AI Models |
| ramalama-login(1) | login to remote registry |
| ramalama-logout(1) | logout from remote registry |
| ramalama-models(1) | list models served by a running inference server |
| ramalama-perplexity(1) | calculate the perplexity value of an AI Model |
| ramalama-pull(1) | pull AI Models from Model registries to local storage |
| ramalama-push(1) | push AI Models from local storage to remote registries |
| ramalama-rag(1) | convert documents to a RAG vector database and package as a container image |
| ramalama-rm(1) | remove AI Models from local storage |
| ramalama-run(1) | run specified AI Model as a chatbot |
| ramalama-sandbox(1) | run an AI agent in a sandbox, backed by a local AI Model |
| ramalama-serve(1) | serve REST API on specified AI Model |
| ramalama-stop(1) | stop named container that is running AI Model |
| ramalama-version(1) | display version of RamaLama |
CONFIGURATION FILES
ramalama.conf (/usr/share/ramalama/ramalama.conf, /etc/ramalama/ramalama.conf, /etc/ramalama/ramalama.conf.d/*.conf, $HOME/.config/ramalama/ramalama.conf, $HOME/.config/ramalama/ramalama.conf.d/*.conf)
RamaLama has built-in defaults for command line options. These defaults can be overridden using the ramalama.conf configuration files.
Distributions ship the /usr/share/ramalama/ramalama.conf file with their default settings. Administrators can override fields in this file by creating the /etc/ramalama/ramalama.conf file. Users can further modify defaults by creating the $HOME/.config/ramalama/ramalama.conf file. RamaLama merges its built-in defaults with specified fields from these files if they exist. Fields specified in the user's file override the administrator's file, which overrides the distribution's file, which overrides the built-in defaults.
RamaLama uses built-in defaults if no ramalama.conf file is found.
If the RAMALAMA_CONFIG environment variable is set, then its value is used for the ramalama.conf file rather than the default.
ENVIRONMENT VARIABLES
RamaLama default behavior can also be overridden via environment variables,
although the recommended way is to use the ramalama.conf file.
| ENV Name | Description |
|---|---|
| HTTP_PROXY, http_proxy | proxy URL for HTTP connections |
| HTTPS_PROXY, https_proxy | proxy URL for HTTPS connections |
| NO_PROXY, no_proxy | comma-separated list of hosts to bypass proxy (e.g., localhost,127.0.0.1,.local) |
| RAMALAMA_CONFIG | specific configuration file to use |
| RAMALAMA_CONTAINER_ENGINE | container engine (Podman/Docker) to use |
| RAMALAMA_FORCE_EMOJI | define whether ramalama run uses emojis |
| RAMALAMA_IMAGE | container image to use for serving AI Models |
| RAMALAMA_IN_CONTAINER | run RamaLama in the default container |
| RAMALAMA_STORE | location to store AI Models |
| RAMALAMA_TRANSPORT | default AI Model transport (huggingface, OCI, ollama) |
| TMPDIR | host temporary directory; used when tempdir is not configured in ramalama.conf(5); defaults to /var/tmp (or ~/.cache/ramalama/tmp in a toolbox) |
See Also
podman(1), docker(1), ramalama.conf(5), ramalama-cuda(7), ramalama-macos(7)
Aug 2024, Originally compiled by Dan Walsh <dwalsh@redhat.com>