dgxllm is a Python package that enables running large LLMs across two NVIDIA DGX Sparks using vLLM for tensor parallelism. It provides a model picker, simple one-command start and stop functionality, and exposes an Anthropic-compatible API endpoint suitable for tools like Claude Code. Designed for developers needing efficient local LLM inference on specialized NVIDIA hardware.
dgxllm is an AI & ML project. It focuses on running large language models efficiently across multiple NVIDIA DGX Spark GPUs with simple controls and compatibility for Claude Code. It is built as an open-source project for developers. The project is open source (MIT). dgxllm is available on the command line, and it can be self-hosted.
dgxllm first shipped in 2026. Among its 5 catalogued features are Model Picker, One-Command Start/Stop, and Anthropic-Compatible API. It exposes integrations via a public API.
Summary written by a language model from the project’s public pages.
What PulseGate has recorded for this listing
Closest matches by what these projects do