sammcj/gollama

Gollama – a terminal UI for managing Ollama models

What it is – Gollama is a small Go program (macOS / Linux) that talks to the local Ollama API and lets you view, run, edit, copy, delete, push and otherwise manipulate the models you have installed. It runs in a text‑based user interface (TUI) that can be navigated entirely with hot‑keys, and it also offers a handful of command‑line flags for quick, script‑able actions.

Why it matters – Ollama is a popular way to serve LLMs locally, but its command‑line tools are limited to basic ollama pull/run etc. Gollama adds a richer, interactive layer: you can sort and filter models by size, quantisation, family, or modification date; see detailed metadata; edit a model’s Modelfile in your favourite editor; estimate the VRAM required for a given quantisation and context length; and push models to a registry – all without leaving the terminal.

Key features (as documented)

• List models with columns for name, size, quantisation, family, modified date, etc.
• Sort and filter on any column.
• Hot‑key actions: Space (select), Enter (run), i (inspect), D (delete), e (edit), c (copy/rename), U (unload all), p/ctrl+p (pull), P (push), t (show running models), etc.
• Edit a model’s Modelfile directly from the UI (uses the editor set in the config or $EDITOR).
• vRAM estimator that prints a table of memory usage for many quantisation levels and context sizes, and can suggest the best quantisation for a given memory budget.
• Simple CLI shortcuts (gollama -l to list, -e <model> to edit, -s <term> to search, --vram … for memory analysis).
• JSON configuration (~/.config/gollama/config.json) for default sort, visible columns, API endpoint, log level, theme, editor, and optional Docker container for running models.
• Theme support (dark‑neon, light‑neon, or custom JSON themes).
• Logging to ~/.config/gollama/gollama.log with configurable level.

Installation – The README recommends the Go toolchain:

go install github.com/sammcj/gollama/v2@latest

Alternatively you can curl the install script or download a pre‑built binary from the releases page and place it on your $PATH.

Typical usage – After installing, just run gollama (or alias it to g). The TUI opens, showing your models. Use the key bindings listed above to manage them. For quick one‑liners you can use the flags, e.g.:

gollama -l                     # list models without UI

gollama -e my‑model               # edit Modelfile of a model

gollama --vram llama3.1:8b‑instruct-q6_K   # show VRAM estimate table

Configuration – A sample config.json is provided; notable fields:

• default_sort / sort_order – how the list is initially ordered.
• columns – which columns to display.
• ollama_api_url – usually http://localhost:11434.
• editor – path to an external editor (VS Code works).
• theme – name of the theme JSON file.
• docker_container – experimental support for running Ollama inside a Docker container.

Status – The project is still under active development (as of the README, v2.0.1 released Dec 2025). The author notes that LM Studio integration has been removed and that development pace has slowed as they focus on other projects.

License – The repository includes a standard open‑source license (the exact SPDX identifier isn’t shown in the excerpt, but it is listed under a License section).

---

All details above are taken directly from the repository’s README; no additional features have been inferred.