# llama-swap

20 Hacker News commenters recommended llama-swap and 1 criticized it, November 2025 to November 2026. It was recommended in 8 of 13 months. It meets the Hacker Approved badge rule.

**Verdict:** People recommend llama-swap to run several local models behind one endpoint, swapping them on demand, often with llama.cpp. One critic says it adds hard-to-debug complexity.

## Why people recommend it

- Works with llama.cpp backends (4 comments): "For software, llama.cpp and llama-swap. GGUF models from HuggingFace. It just works." ([source](https://news.ycombinator.com/item?id=46349175)) "I just switched from Llama.cpp to Llama swap with the help of codex. It was great." ([source](https://news.ycombinator.com/item?id=48510147)) "Currently relies on llama-swap / llama.cpp as the underlying inference engine" ([source](https://news.ycombinator.com/item?id=48628843))
- Serves many models from one endpoint (5 comments): "allows running multiple models on a single endpoint." ([source](https://news.ycombinator.com/item?id=48544986)) "llama-swap which allows you to run one frontend which provides a list of models" ([source](https://news.ycombinator.com/item?id=47214054)) "so I use llama-swap to dynamically swap models" ([source](https://news.ycombinator.com/item?id=48557176))
- Per-model llama.cpp configuration (2 comments): "its configuration allows me to tailor llama.cpp settings per-model" ([source](https://news.ycombinator.com/item?id=47821846)) "my llama.cpp + llama-swap setup switches to a different configuration that reloads the same model" ([source](https://news.ycombinator.com/item?id=48141760))
- Better than Ollama, Ollama-like feel (2 comments): "llama.cpp and llama-swap do this better than Ollama" ([source](https://news.ycombinator.com/item?id=47505927)) "I use llama-swap to emulate the Ollama experience." ([source](https://news.ycombinator.com/item?id=48546870))

## Cite as

Hacker Approved, "llama-swap", https://hackerapproved.com/p/llama-swap/ (counts from 1,795,640 Hacker News comments, November 2025 to November 2026).