PrivateGPT is an open-source API layer that turns local models into production AI applications, following the Claude API model. It provides a messages API, file and artifact ingestion, retrieval with citations, embeddings and orchestration, and a built-in workbench UI for testing and demos.
LocalAI is described as an open source AI engine that runs models, including LLMs, vision, voice, image and video, on any hardware without a GPU. It presents itself as a drop-in replacement for most upstream APIs and includes a built-in agent platform powered by LocalAGI, whose agents can reason, use tools and maintain memory. Image generation with Stable diffusion on CPU is also supported.
KoboldAI presents tools for local AI use. KoboldCPP is a local LLM API server for driving a backend, and KoboldAI Lite is a lightweight interface for accessing AI API endpoints. KoboldAI.net delivers KoboldAI Lite as a free web service. Downloads for KoboldCPP are offered for Windows, Linux and Mac, and community support is available on Discord.
Leon is an open-source personal assistant that can run on a user's server. It is built on Node.js, Python and artificial intelligence concepts, and responds to requests. It uses a modular architecture for creating or using skills, and supports several cloud and offline text-to-speech and speech-to-text solutions.
Miku.gg is an open source chatbot platform for running bots. Users can download the source code and run their own chatbots or use the live demo. It works with any OpenAI-like APIs and currently supports Aphrodite, TextGen WebUI and Tabby. Bots can show emotion through images and use TTS, with Azure TTS, NovelAI and ElevenLabs supported. A bot directory allows sharing, downloading or trying bots without local installation.
Vicuna-13B is an open-source chatbot trained by fine-tuning LLaMA on user-shared conversations collected from ShareGPT. Its code, weights and an online demo are publicly available for non-commercial use, and training cost around $300. The team reports a preliminary GPT-4-judged evaluation, described as non-scientific, showing more than 90% quality of ChatGPT and Bard.
A links page listing resources related to running local language models. It gathers guides, including a quick start guide for running models locally, model sources such as HuggingFace for current quants, and inferencing frontends including Text Gen WebUI.
Community notes
Be the first to share how this project worked for you.
Sign in with a verified account to leave a review or join the discussion.
Sign in to reviewKeep exploring
LocalAI is described as an open source AI engine that runs models, including LLMs, vision, voice, image and video, on any hardware without a GPU. It presents itself as a drop-in replacement for most upstream APIs and includes a built-in agent platform powered by LocalAGI, whose agents can reason, use tools and maintain memory. Image generation with Stable diffusion on CPU is also supported.
KoboldAI presents tools for local AI use. KoboldCPP is a local LLM API server for driving a backend, and KoboldAI Lite is a lightweight interface for accessing AI API endpoints. KoboldAI.net delivers KoboldAI Lite as a free web service. Downloads for KoboldCPP are offered for Windows, Linux and Mac, and community support is available on Discord.
Leon is an open-source personal assistant that can run on a user's server. It is built on Node.js, Python and artificial intelligence concepts, and responds to requests. It uses a modular architecture for creating or using skills, and supports several cloud and offline text-to-speech and speech-to-text solutions.
Miku.gg is an open source chatbot platform for running bots. Users can download the source code and run their own chatbots or use the live demo. It works with any OpenAI-like APIs and currently supports Aphrodite, TextGen WebUI and Tabby. Bots can show emotion through images and use TTS, with Azure TTS, NovelAI and ElevenLabs supported. A bot directory allows sharing, downloading or trying bots without local installation.
Vicuna-13B is an open-source chatbot trained by fine-tuning LLaMA on user-shared conversations collected from ShareGPT. Its code, weights and an online demo are publicly available for non-commercial use, and training cost around $300. The team reports a preliminary GPT-4-judged evaluation, described as non-scientific, showing more than 90% quality of ChatGPT and Bard.
A links page listing resources related to running local language models. It gathers guides, including a quick start guide for running models locally, model sources such as HuggingFace for current quants, and inferencing frontends including Text Gen WebUI.
