otty / apps
All apps/RamaLama

RamaLama

ramalama

A tool for managing and serving AI models using containers.

Runs as ramalama|macOS, Linux, Windows|ramalama.ai
Open form in Otty otty.toml

Command form

ai-draft · 2 tasks · 15 fields

Chat with a model or serve it over HTTP in a container. Image, device, RAG, speculative-decoding and TLS options are left out.

Opens the form; the finished command lands at your prompt, unrun
Run in Otty
otty gui ramalama

ramalama run

Run a model

Chat with a model interactively.

Run a modelExample values, fixed in this page — nothing here reaches a terminal
Basic
Model*
string
GPU backend
8 choicesenum
Max tokens
integer
TasksEach one presets the form for one job
Run a model
modelbackendmax_tokens
ParametersOnly these flags are editable in the form
Model
string · positional 0
GPU backend
enum
--backendautovulkanrocmcudasyclopenvinocannmusa
Container name
string
--name-n
Context size
integer
--ctx-size-c
Temperature
number
--temp
CPU threads
integer
--threads-t
Max tokens
integer
--max-tokens

ramalama serve

Serve a model

Serve a model on a local port.

Serve a modelExample values, fixed in this page — nothing here reaches a terminal
Basic
Model*
string
GPU backend
8 choicesenum
Port
integer
Run in background
--detach
TasksEach one presets the form for one job
Serve a model
modelbackendportdetach
ParametersOnly these flags are editable in the form
Model
string · positional 0
GPU backend
enum
--backendautovulkanrocmcudasyclopenvinocannmusa
Container name
string
--name-n
Context size
integer
--ctx-size-c
Temperature
number
--temp
CPU threads
integer
--threads-t
Port
integer
--port-p
Run in background
boolean
--detach-d
SourcesEvery flag above is checked against these
man
ramalama run [options] model [arg ...] --backend=auto | vulkan | rocm | cuda | sycl | openvino | cann | musa --ctx-size, -c size of the prompt context --max-tokens=integer Maximum number of tokens to generate. Set to 0 for unlimited output (default: 0). --name, -n Name of the container to run the Model in. --temp="0.8" Temperature of the response fro…
manualmain 2026-10-27

github.com

man
ramalama serve [options] [model ...] --backend=auto | vulkan | rocm | cuda | sycl | openvino | cann | musa --ctx-size, -c --detach, -d --name, -n --port, -p --temp="0.8" --threads, -t ramalama serve -d -p 8080 --name mymodel ollama://smollm:135m
manualmain 2026-10-27

github.com

Before it fills your prompt, Otty checks the installed ramalama against its version. A form that does not match your version is not offered.