llama-swap
Hacker ApprovedModel proxy in AI tools
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.
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.
21 people gave an opinion
- recommend it
- 20
- criticize it
- 1
- approval share
- 95%
Recommended in 8 of 13 months
- November 2025: not recommended
- December 2025: recommended
- January 2026: recommended
- February 2026: not recommended
- March 2026: recommended
- April 2026: recommended
- May 2026: recommended
- June 2026: recommended
- July 2026: not recommended
- August 2026: recommended
- September 2026: recommended
- October 2026: not recommended
- November 2026: not recommended (month in progress)
Why it carries the badge
- 20 people recommend itNeeds 10 or more, met.
- 20 recommenders for every criticNeeds 2 to 1 or better, met.
- Recommended in 8 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Works with llama.cpp backends
“For software, llama.cpp and llama-swap. GGUF models from HuggingFace. It just works.”
Comment from 21 Dec 2025 in “A guide to local coding models” “I just switched from Llama.cpp to Llama swap with the help of codex. It was great.”
Comment from 12 Jun 2026 in “Kimi K2.7-Code: open-source coding model with better token efficiency” “Currently relies on llama-swap / llama.cpp as the underlying inference engine”
Comment from 22 Jun 2026 in “Llama-dash – One go-to control plane for local inference” - Comment from 14 May 2026 in “What's in a GGUF, besides the weights – and what's still missing?”
Serves many models from one endpoint
“allows running multiple models on a single endpoint.”
Comment from 15 Jun 2026 in “Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?” “llama-swap which allows you to run one frontend which provides a list of models”
Comment from 2 Mar 2026 in “Qwen3.5 122B and 35B models offer Sonnet 4.5 performance on local computers” “so I use llama-swap to dynamically swap models”
Comment from 16 Jun 2026 in “Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?” - Comment from 20 Jun 2026 in “My self-hosted local LLM server setup”
- Comment from 12 Aug 2026 in “llama.cpp”
Per-model llama.cpp configuration
“its configuration allows me to tailor llama.cpp settings per-model”
Comment from 19 Apr 2026 in “Stop Using Ollama” “my llama.cpp + llama-swap setup switches to a different configuration that reloads the same model”
Comment from 14 May 2026 in “What's in a GGUF, besides the weights – and what's still missing?” Better than Ollama, Ollama-like feel
“llama.cpp and llama-swap do this better than Ollama”
Comment from 24 Mar 2026 in “Run a 1T parameter model on a 32gb Mac by streaming tensors from NVMe” “I use llama-swap to emulate the Ollama experience.”
Comment from 15 Jun 2026 in “Ask HN: Has anyone replaced Claude/GPT with a local model for daily coding?”
What people said
23 of the 26 comments that mention it, newest first. Each links to the original comment.
“Just start llama-swap with the models i have downloaded, add llama-swap as a provider in bifrost”
“I think that's where the sweet spot is right now.”
“llama-swap can easily eject both models if necessary”
“Adding llama-swap adds a fair amount of complexity that is unpleasant to debug”
“there's still benefits if you're using 24-32GB cards that can run multiple models simultaneously”
“make this pretty bullet proof and offer an OAI style API and compatibility for local Agents”
“It's also convenient if you want to run separate builds behind a single config”
“llama-swap provides a nice UI for monitoring performance and logs, and even the ability to stop an infinite session that consumes GPU resources”
“llama-swap was still the go-to solution”
“it can manage arbitrary backends, not just llama-server instances”
“llama-swap, if you want to use other models too”
“Currently relies on llama-swap / llama.cpp as the underlying inference engine”
“Useful when juggling models on one box.”
“so I use llama-swap to dynamically swap models”
“I use llama-swap to emulate the Ollama experience.”
“allows running multiple models on a single endpoint.”
“I just switched from Llama.cpp to Llama swap with the help of codex. It was great.”
“my llama.cpp + llama-swap setup switches to a different configuration that reloads the same model”
“I've got llama-swap running and set up on my tailnet as a tailscale service”
“its configuration allows me to tailor llama.cpp settings per-model”
“llama.cpp and llama-swap do this better than Ollama”
“llama-swap which allows you to run one frontend which provides a list of models”
“For software, llama.cpp and llama-swap. GGUF models from HuggingFace. It just works.”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.