Open-source ComfyUI node offloads text encoders to a spare GPU on another machine
randomlyfixed · reddit · 2026-11-16
Developer randomlyfixed released comfyui-remote-text-encoder, an open-source custom node that moves ComfyUI's CLIP text encoder off your rendering GPU and onto another machine's idle GPU.
- How it works: install the node pack on both machines, run one as a server (toggle in settings or via CLI arg), and swap the clip loader in your workflow for the remote clip loader.
- The server loads models on demand from the default text-encoder directory; the client can auto-detect diffusion model node types and route requests accordingly.
- Extras include model pinning, model capability scanning, and support for dual GPUs on a single machine.
More from Infra
- Inside SGLang Summit 2026: 23 sessions in one track and a striking open-source turnout — ying11231 · 2026-11-16
- EPYC 32-core + 512GB DDR4 + 44GB VRAM: is a DIY local LLM rig still worth it? — sitytitan · 2026-11-16
- Forensic Analysis: Top Chinese AI Models Fingerprinted to Export-Banned NVIDIA Blackwell GPUs — elie · 2026-11-16
- Samsung open-sources LittleBit: 13B LLM under 1GB with sub-bit weights and XOR compute — dr_cintas · 2026-11-16
- Samsung to invest $1.9B in Austin to make chips for SpaceX Starlink satellites — elonmusk · 2026-11-16
- Jensen Huang rumored plotting return to Nvidia CEO as Musk, Sweeney back public letter — The Verge AI · 2026-11-16