Parallax
Visit ToolParallax is a distributed model serving framework that lets you build your own AI cluster anywhere. It supports hosting local LLMs on personal devices and offers cross-platform compatibility.
At a glance
Trending
Parallax is a distributed model serving framework that lets you build your own AI cluster anywhere. It supports hosting local LLMs on personal devices and offers cross-platform compatibility.
Trending
About
Parallax is an open-source, fully decentralized inference engine developed by Gradient, designed to enable users to build their own AI clusters for model inference. It allows for the deployment of models onto a set of distributed nodes, regardless of their varying configurations and physical locations. Key features include the ability to host local LLMs on personal devices, cross-platform support, pipeline parallel model sharding, and paged KV cache management with continuous batching for Mac. The framework also provides dynamic request scheduling and routing for high performance, utilizing P2P communication powered by Lattica and GPU backend powered by SGLang and vLLM, with MLX LM for Mac backend. It supports a wide range of models from providers like HuggingFace, DeepSeek, MiniMax AI, Z AI, Moonshot AI, Qwen, OpenAI, and Meta.
Capabilities
Pricing & Plans
Open Source
Free
FAQs
Trending