AI JIGYASAEXPLORE•CREATE•LEARN•LOCAL AI
SUPPORT

Home / News

What's
new.

New models, new nodes from the community, and releases worth your GPU time — each with the source link, and a workflow when there is one.

Queue History
stories 12
latest Sep 23, 2026

Fresh out of
the sampler.

LATEST · ARTICLE
SEP 23, 2026

Introducing Comfy Router: One API for Frontier Media Models

Comfy Router offers a unified API for frontier media models. Developers build once and add models as they launch, and users choose where jobs run.

Comfy blog
READ THE STORY
  1. #011 · ARTICLE

    MiniMax H3 Video VAE Speed Boost

    MiniMax H3 video VAE now processes video data significantly faster. Encodes up to 2.2x quicker and decodes 1.4-2.7x faster.

    Comfy blog
  2. #010 · LORA

    Qwen-Image-2.1-viggle-turbo LoRA for text-to-image

    A distilled version of Qwen-Image-2.1 with 4-step text-to-image capability. License restricts commercial use.

    Viggle on Hugging Face
  3. #009 · NODE

    ComfyUI node speeds up QwenImage2.1 inference with TE-Speed-QwenImage21

    A new ComfyUI node accelerates QwenImage2.1 inference. It uses output prediction to reduce processing time.

    tl2012tl on GitHub
  4. #008 · NODE

    Comfyui-Spectrum-Qwen2.1 gains 47 stars

    A new ComfyUI node for Qwen2.1 model integration has received 47 stars on GitHub.

    awdqwdasdg on GitHub
  5. #007 · RELEASE

    ComfyUI v0.37.0 release adds new features and fixes

    ComfyUI v0.37.0 introduces partner node updates and performance improvements.

    Comfy-Org on GitHub
  6. #006 · MODEL

    Supra2-IMG model achieves SOTA quality with 100M parameters

    Supra2-IMG is a text-to-image model with 100M parameters trained on 5.6M images.

    SupraLabs on Hugging Face
  7. #005 · NODE

    ComfyUI-Qwen-Image-2.1-Prompt-Enhancer adds T2I and Edit nodes

    A new ComfyUI node for Qwen-Image-2.1 prompt enhancement has 117 stars. It supports text-to-image and image editing.

    benjiyaya on GitHub
  8. #004 · MODEL

    Qwen-Image-2.1-PE-T2I prompt rewriting model released

    A new prompt rewriting model for Qwen-Image-2.1 enables text-to-image generation with language flexibility. It outputs rewritten prompts and aspect ratios.

    Qwen on Hugging Face
  9. #003 · NODE

    TE-Speed-VOSR2 speeds up ComfyUI image and video upscaling

    New ComfyUI node accelerates VOSR2 image and video upscaling with memory optimizations.

    tl2012tl on GitHub
  10. #002 · RELEASE

    ComfyUI v0.36.0 adds model blueprints, video concatenation and Marigold v2

    New model blueprints, video features, and partner node additions in ComfyUI v0.36.0.

    Comfy-Org on GitHub
  11. #001 · NODE

    AIFISHER connects ComfyUI to AI models via local canvas

    AIFISHER provides a local canvas for connecting AI models with ComfyUI and RunningHub. It's open source but restricted to non-commercial use.

    Work-Fisher on GitHub

Spotted something new?

A model, a node, a release worth covering — send it over.

SEND A TIP