November 2025 Summaries

The finale of Comfy Challenge Season 1 celebrates by transforming the historic Niš Fortress in Serbia into a projection canvas for 15 animations created using ComfyUI, marking a collaboration with the University of Niš and the Festival of Light. Participants are tasked with creating videos themed "Echoes of Time," exploring the convergence of past, present, and future, which will be projected onto the fortress walls in December using a 30,000-lumen projector, accompanied by sound and video documentation. The top submissions will receive prizes, including a free Comfy Cloud account and a cash reward for the top winner. Participants must adhere to specific animation requirements to ensure accurate projection mapping, with the University of Niš testing selected entries on a 3D-printed scale model of the fortress.

Topaz Labs, known for its industry-leading AI enhancement tools, has partnered with ComfyUI to integrate its advanced image and video enhancement capabilities directly into the platform. This collaboration introduces several innovative models such as Starlight Fast and Astra for video upscale, Apollo for frame interpolation, and Bloom for image enhancement, allowing users to upscale videos to 4K and images to 8K. These tools offer features like subject detection, face enhancement, and feature preservation, providing professional-grade improvements while maintaining the original aesthetics. Users can easily access these features by updating ComfyUI or using Comfy Cloud, facilitating the creation of high-quality visual content with enhanced detail and resolution.

Z-Image Turbo, developed by Alibaba Tongyi Lab, is a highly efficient 6B-parameter image generation model designed for creating hyper-realistic portraits with exceptional detail fidelity. Its distilled version is optimized for consumer GPUs with 16GB of VRAM, allowing it to maintain high performance while being accessible to a broader range of users. The model excels in producing photorealistic images with strong detail and is part of a broader initiative to release the full, non-distilled foundation model for community-driven fine-tuning and development within the open-source ecosystem. Users can explore its capabilities through ComfyUI, with example workflows demonstrating its ability to generate high-quality, cinematic images with varied lighting and texture effects.

FLUX.2 is a newly open-sourced, next-generation image model from Black Forest Labs that is now supported in ComfyUI, offering photorealistic outputs up to 4MP with enhanced lighting, skin, fabric, and hand detail. This model emphasizes visual intelligence over mere image synthesis, providing reliable multi-reference consistency and professional-grade control, making it ideal for applications like ad variants, product shots, and editorial series. FLUX.2 is available in different versions, including open-source models for local use and an API version for broader accessibility, seamlessly integrating into Comfy Cloud and local ComfyUI setups. It features high-resolution photorealism, precise text rendering, and structured prompting for programmatic workflows, suitable for environments such as e-commerce, marketing, and social content creation. Users can run FLUX.2 locally or through Partner Nodes, offering flexibility for different hardware configurations and workflow requirements.

Comfy Cloud is introducing new features and a revised pricing model to enhance user experience and resource efficiency. The platform has upgraded to Blackwell RTX 6000 Pro GPUs, which offer double the speed of A100s, and increased memory capacity, facilitating more complex workflows like video upscaling without additional costs to users. Starting December 8th, 2025, users can upload their own LoRAs and benefit from extended workflow durations under the new Creator and Pro plans, respectively. Comfy Cloud is also implementing a unified credit system called Comfy Credits, which consolidates all platform expenditures under a single balance, allowing for usage-based billing to maintain access to high-quality GPUs without increasing base prices. Existing subscribers will transition to a new Standard subscription model, receiving a 30% credit bonus as part of a Founder's edition benefit, ensuring equitable access and the continuous introduction of premium features.

HunyuanVideo 1.5, developed by Tencent's Hunyuan team, is a lightweight yet powerful 8.3 billion parameter model that facilitates high-quality video generation on consumer GPUs with 24GB VRAM, significantly lowering entry barriers without sacrificing quality. This model supports advanced Text-to-Video and Image-to-Video generation, offering precise control over elements like camera movements, physics, and emotional expressions, delivering native 720p output that can be upscaled to 1080p with professional aesthetics. Its versatility allows for a wide array of styles, including realistic, anime, and 3D, as well as in-video text rendering in both Chinese and English. The model's integration into ComfyUI enables users to update their local versions and utilize detailed workflows to create cinematic-quality videos with ease.

Nano Banana Pro, Google DeepMind's new high-fidelity image generation and editing model, is now available in ComfyUI, offering users production-ready visuals with advanced features. This flagship model draws on Google's knowledge base to generate accurate, real-world images, supports clean text rendering and translation across ten languages, and provides high-resolution images with studio controls, such as blending, angle adjustment, and color grading in native 4K. It excels in consistency, transforming sketches into detailed visuals. Users can access Nano Banana Pro via Comfy Cloud without additional setup, while local users are advised to update ComfyUI for integration. The model's pricing is $0.134 per 1K & 2K image output and $0.24 per 4K image output, and it supports multi-stage creation by chaining Nano Banana Pro nodes or Partner Nodes. Additionally, the integration of Google’s latest language model, Gemini 3, facilitates image captioning and prompt generation, enhancing the workflow for ComfyUI users.

Comfy Challenge #9, themed "Cloudcrafted," invites participants to create conceptual videos inspired by clouds, alongside announcing the winners from the previous "Spooky Renders" challenge. The winners of the previous challenge, including a short film by Nino - Soda Zitron Media, and top submissions by Stefan Rademakers & Jonathan Kray, Saheer Muhammed SM, and Anton Lisitskii, received Halloween-themed prizes. For the current challenge, participants can win $100 cash or $200 in ComfyUI credits, special merchandise, and a 1-month free Comfy Cloud experience for 20 randomly selected entries. The videos should be submitted by November 10, 7 PM PST, and must be under 30 seconds, in 1:1 format, and over 720P, while being free from watermarks and minimal in captions. The blog encourages creativity and provides an opportunity to explore the new Comfy Cloud Public Beta.

Comfy Cloud has launched its public beta phase, expanding access beyond the previous waitlist and inviting user feedback to enhance its offerings. This cloud-based platform extends the capabilities of ComfyUI directly to users' browsers, supported by NVIDIA A100 GPUs and granting access to over 400 open-source models and multiple community-built extensions. The service is priced at $20 per month, including $10 in credits for partner nodes and fair-use GPU access, with a future pricing model that charges only for active workflow runtime. During the beta, limitations include a maximum of 30 minutes per workflow and a single queued workflow, although higher limits are available for heavier users upon request. Comfy Cloud aims to cater to creators, studios, and developers, with plans to introduce more custom nodes, additional GPU options, API deployment, parallel workflow execution, and team collaboration features. User feedback is encouraged through the ComfyUI Discord to help guide the platform's development.