gollama mit Homebrew, Nix installieren

Projektlinks

• Homepage
• Repository
• Upstream-Dokumentation
• Paketmanager-Seite

Befehle und Aliase

• gollama

Projektgeschichte

Sam McLeod's June 2024 project post says Gollama started as a rewrite of `llamalink`, then expanded into a friendlier model-management interface. Early documentation emphasized linking Ollama models to LM Studio as well as listing and deleting local models.

The README later broadened the tool with Modelfile editing, registry push/pull operations, running-model views, vRAM estimation, search, and configurable Ollama API host options.

The maintainer documented a 2025 shift: LM Studio linking was removed in the v2.0.1 era because ongoing upstream changes and local-configuration differences made the feature costly to maintain, and development slowed as the maintainer's own model-serving workflow moved away from Ollama.

Adoptionsgeschichte

Gollama belongs to the local-LLM wave around Ollama: it is a companion tool for users who accumulate many local model tags and need a terminal interface to clean, inspect, run, and organize them. The batch input records Homebrew and Nix packages, matching the macOS/Linux developer audience named in the README.

Its adoption story is narrower than Ollama itself but meaningful for package catalogs because it packages local model housekeeping into one Go binary and can be installed without a desktop UI.

Wie es verwendet wird

The primary workflow is running `gollama` to open the TUI, then using hotkeys to select, run, inspect, delete, edit, copy, pull, push, sort, and unload models. The CLI also supports list, search, edit, host override, custom Ollama directory, log-level, and vRAM estimation flags.

Gollama reads and writes user configuration under `~/.config/gollama/`, including sorting, displayed columns, API URL, log level, and related display options. It may store an Ollama API key in that configuration when users choose to configure one, but the package itself is not primarily a credential manager.

Warum Paket-Nerds sich dafür interessieren

For package nerds, Gollama is a good example of the post-Ollama helper-binary ecosystem: small Go tools that wrap a local AI runtime with ergonomic terminal workflows. It is easy for Homebrew and Nix users to install beside Ollama, LM Studio, llama.cpp, and model-management scripts.

The v2 removal of LM Studio linking is also significant: it shows how fast-moving local-AI tools force package maintainers and users to track feature removals, not just version bumps.

Zeitleiste

• 2024: Project post announces Gollama as a Go-based Ollama model manager and explains its origin as a rewrite of llamalink.
• 2024: README describes TUI model listing, filtering, sorting, deleting, inspecting, and LM Studio linking.
• 2025: Release notes and README document removal of LM Studio linking in the v2 line.
• 2025-2026: README emphasizes model-management, registry, running-model, and vRAM-estimation workflows.

Related projects

• Ollama is the local model runtime Gollama manages.
• llamalink is the maintainer's earlier project that Gollama rewrote and expanded.
• LM Studio, llama.cpp, and llama-swap are named in maintainer documentation around the removed linking feature and the maintainer's model-serving workflow.

Quellen

• https://github.com/sammcj/gollama
• https://github.com/sammcj/gollama/releases
• https://github.com/sammcj/llamalink
• https://ollama.com/
• https://smcleod.net/2024/06/gollama-ollama-model-manager/