Hi Product Hunt — OpenMake is not a company. We are a small team that enjoys open-source software, open hardware, and running services on machines we control. We built OpenMake LLM because local-model support often feels bolted onto cloud-first products. We wanted local and open-weight models to be the normal path, with every external provider remaining an explicit BYOK choice. Our real development setup is a Mac mini running OpenMake, PM2, and a loopback-only LiteLLM gateway, paired with an NVIDIA DGX Spark powered by the GB10 Grace Blackwell Superchip. The DGX Spark serves private Qwen, BGE, and FLUX vLLM endpoints over Tailscale. This is our test setup, not a hardware requirement; any compatible OpenAI-style endpoint can replace it. Normal chat resolves one model and shares one provider-gated MCP tool loop. Multi-agent Discussion is an explicit mode rather than something invoked for every prompt. Autonomous tasks run shell, Python, browser, and file tools inside persistent Docker sandboxes with checkpoints, manual resume, and human approval boundaries. Deep Research keeps citations, and results can become HTML, PDF, or DOCX artifacts. OpenMake LLM is MIT licensed and still not a one-click lightweight install. Today it needs Node.js 24, Docker, PostgreSQL, and an OpenAI-compatible model endpoint. Local vLLM also needs enough compatible GPU memory. We would especially value feedback on installation, working GPU/model combinations, role-based model routing, and which agent actions should always require human approval. Source and real screenshots: https://github.com/openmake/open... Website: https://openmake.cc/en/?utm_sour... Live demo: https://chat.openmake.cc

OpenMake LLM

Local-first AI workspace for your models and tools

Local-first AI workspace for your models and tools