About a dozen open tools released around MOSS-Image-SFT in one week

From October 22 to 31, 2026, the community released about ten tools for the open-source MOSS-Image-SFT model, including the Edit-HQ edit-LoRA, the PiD-SuperFast-HighRes pipeline with the NVIDIA PiD pixel decoder, and the moss4gb 4-bit build.

Compiled by Sergey KostenchukPublished 2026-10-31Updated 2026-10-31

2026-10-31 Multimodal NVIDIA

🎨 MOSS-Image-SFT has been surrounded by a dozen open tools

From October 22 to 31, the following were released around the MOSS-Image-SFT model (5B DiT, Qwen3.5-9B encoder, Flux2 VAE): the Edit-HQ instruction edit-LoRA (armV5 checkpoint, 50 steps, ~20+ GB VRAM), the PiD-SuperFast-HighRes pipeline with the NVIDIA PiD decoder up to 2816x1536 in 4 steps, the moss4gb 4-bit build for 4 GB GPUs, a transparent background LoRA, pixel art with the pxl8x trigger, and PixText. All links have been verified.

🌍 The base has turned into an open platform: generation and editing quality and speed are being enhanced through LoRAs and pipelines without fine-tuning the base model — competition has shifted to this layer.

👤 You can already build a workflow for your specific GPU: from 4 GB (moss4gb, rendering from 256x256) to 20+ GB (Edit-HQ), fast 4K rendering via PiD, and transparent backgrounds via transparent-lora.

Source 1: https://huggingface.co/akhaliq/MOSS-Image-Edit-HQ Source 2: https://huggingface.co/lvladikov/MOSS-PiD-SuperFast-HighRes

Sources