llama.cpp
Generative Artificial Intelligence (GenAI)Fast C/C++ engine that runs open AI models on ordinary CPUs and GPUs, with a built-in chat page and OpenAI-style API
- Stars
- ★ 132k
- Licence
- MIT
- Updated
- Docker
Alternative to ChatGPT, OpenAI API
Self-hosted alternatives to OpenAI API
Last updated · 6 options
Quick answer: llama.cpp has the most stars on its repository of the 6 options here (★ 132k), followed by vLLM (★ 94.1k). Stars show popularity, not fit, so check formats, platforms and licence for your own work.
OpenAI's paid API for GPT models, used by apps and developers. Here are 6 self-hosted options, most-starred first.
vLLM
Generative Artificial Intelligence (GenAI)High-throughput server for running open language models on GPUs behind an OpenAI-compatible API
- Stars
- ★ 94.1k
- Licence
- Apache-2.0
- Updated
- Docker
Alternative to OpenAI API
LiteLLM
Generative Artificial Intelligence (GenAI)Self-hosted AI gateway that puts 100+ model providers behind one OpenAI-style API
- Stars
- ★ 61.6k
- Licence
- MIT
- Updated
- Docker
Alternative to OpenAI API
SGLang
Generative Artificial Intelligence (GenAI)Fast GPU serving framework for large language and vision models
- Stars
- ★ 37.4k
- Licence
- Apache-2.0
- Updated
- Docker
Alternative to OpenAI API
Speaches
Generative Artificial Intelligence (GenAI)OpenAI-compatible speech server for transcription and text-to-speech
- Stars
- ★ 3.7k
- Licence
- MIT
- Updated
- Docker
Alternative to OpenAI API, ElevenLabs
RamaLama
Generative Artificial Intelligence (GenAI)Runs AI models in Podman or Docker containers with one command
- Stars
- ★ 3.1k
- Licence
- MIT
- Updated
- Docker
Alternative to OpenAI API
No tools match these filters. Suggest one.
The ARM filter only shows apps whose own docs say they run on ARM or a Raspberry Pi. Others may work too.
FAQ
Is there a self-hosted alternative to OpenAI API?
Yes. SelfHostAlts lists 6 options: llama.cpp, vLLM, LiteLLM, SGLang, Speaches and RamaLama.
Which OpenAI API alternatives are open source?
llama.cpp (MIT), vLLM (Apache-2.0), LiteLLM (MIT), SGLang (Apache-2.0), Speaches (MIT) and RamaLama (MIT).
Which OpenAI API alternatives can run with Docker?
llama.cpp, vLLM, LiteLLM, SGLang, Speaches and RamaLama (all of them).
Related categories: Generative Artificial Intelligence (GenAI)
Self-hosted apps are rarely drop-in replacements. Check the import options, mobile apps and requirements before you move your data, and keep backups.