Release Notes
23 min
What's New
- ComfyUI Frontend upgraded to v1.42.15 with major Nodes 2.0 fixes.
- Unified File Uploader is now live across the entire platform (File Explorer, Folder Upload, Node Uploads). Supports files up to 5 TB.
- Floyo API expanded with workflow discovery endpoints — search, filter, sort, and paginate workflows programmatically. Retrieve full workflow details including prompt JSON, nodes, models, and stats. See documentation →
- Learning Academy updated with ComfyUI 101 course videos, now available in Floyo.
Models & Nodes
- AnimaLLLiteApply (ComfyUI-Anima-LLLite)
- StyleModelLoaderAny (ComfyUI-LoaderUtils)
- Floyo Sticky Note
- ComfyUI-Fluxtapoz
- ComfyUI-FramePackWrapper
- NNLatentUpscale
- SeedVarianceEnhancer
- comfyui-klein-edit-composite
- ComfyUI-Multi-Analysis-Heatmaps
- ComfyUI-FailSafe-Translate-Node
- ComfyUI-PiD
- ComfyUI-TextureAlchemy
- ComfyUI-RH-Bernini
- ComfyUI-CoCoTools
- DerfuuComfyUIModdedNodes
- ComfyUI-EasyColorCorrector
- Irodori TTS (comfy-Irodori-TTS)
Platform Improvements / Bugfixes
- Nodes 2.0: resolved all known layout issues with node sizing and positioning. Fixed subgraph workflow execution. Fixed connector line flickering. Note: nodes in Nodes 2.0 are slightly larger than Nodes 1.0 by design — older published workflows may appear slightly more compact.
- Pricing page: Free Trial Flex FloTime updated to 15 minutes (was 5 minutes).
- Node detail page improvements.
What's New
We're introducing Verified Private and Commercial Use tags so you can quickly see how each model handles privacy and licensing.
Why this matters...
With the large number of AI models available, you may find it difficult to keep track and make sense of all the model licensing rules particularly around two key concerns:
- Are my assets safe and private from being tracked and from being trained on
- Can I freely use the generated content commercially in my business, products and services
Floyo makes it easy to see whether a model is Verified Private and cleared for Commercial Use. We assess each model and tag it in the UI so you can identify compliant options at a glance. You can also block Unverified Private or Non-Commercial Use models — or specific models — for your entire team, so no one uses them by accident.
Verified Private
Models that the provider does not track or train on your data.
Commercial Use
Models with outputs cleared for commercial production.
Verified Private and Commercial Use Tags
Tags appear on workflow pages and on applicable custom nodes in the App canvas:
Block Models for your Team
You can prevent your team from using models that don't meet your privacy or licensing requirements. See Manage Models & Nodes (OLD)Manage Models for step-by-step instructions.
How enforcement works
Blocked models are excluded from loading into the app canvas and any workflow run with a blocked model, including Floyo API runs, will fail accordingly.
What’s New
- ComfyUI upgraded to v0.19.1, unlocking support for newer models like Flux.2 Klein 9B KV and Ernie.
- VHS node updated and now fully compatible with Nodes 2.0 workflows.
- Verified 3D workflows now fully working.
- Continued improvements to serverless infrastructure with idle policies and better scaling.
New Workflows, you can try
- ERNIE Image, Text to Image: Generate images with Baidu's ERNIE Image model. Write a short prompt and let the built-in AI enhancer expand it into rich detail.
- Seedance 2.0
- Product Tutorial Videos in Seconds: Turn any product photo into a realistic UGC-style product review and tutorial video.
_1777212190508.webp?width=1400&height=620&quality=80&resize=cover "Document image")
- AI UGC Fashion & Accessory Videos: Create viral UGC-style fashion and accessory videos from just two photos.
- AI Virtual Try-On Fashion Videos: The workflow wears your outfit, moves naturally, talks to camera, and shows the full look — just like a real influencer video.
- Make ASMR Unboxing Videos: Turn your product image and a style reference photo into a cinematic unboxing video.
Platform Improvements/Bugfixes
- Flux.2 Klein 9B KV: Advanced image generation and editing workflows.
- App Guided tour for new users.
- Native 3D node are now fully compatible on Floyo.
What’s New
- Qwen3-ASR: Powerful all-in-one speech recognition models that support language identification and ASR for 52 languages and dialects
- IndexTTS2: Industrial-Level Controllable and Efficient Zero-Shot Text-To-Speech System.
- New Seedance 2.0, Mimic Motion, and Depth AI capabilities for motion, lighting, and 3D-aware edits
- Advanced image upscaling, tiling, and restoration pipelines now supported (HyperTile and RealRestorer)
- Panorama, 3D, and animation workflows unlocked (Anima3, Panorama Stickers, motion pipelines)
- Serverless performance & pricing improvements.
New Workflows, you can try
- Seedance 2.0
- Seedance 2.0 Reference-to-Video: Upload up to three images, write a prompt that describes what moves and how, and the model builds a video from them.
- Seedance 2.0 Fast - Image to Video with Audio: Animate any image with built-in audio generation, start and end frame control, and multiple aspect ratios.
- Seedance 2.0 - Image to Video: Animate any image with start and end frame control, and clips up to 10 seconds.
- Seedance 2.0 Fast - Text to Video: Describe a scene, pick a duration and aspect ratio, get a clip with synced sound.
_1776068409140.webp "Document image")
- IndexTTS2
- IndexTTS2 Voice Cloning with Emotion Control: Upload a short reference clip of the voice you want to clone, type your script, and run. The output is a high-quality WAV file in that voice.
- Qwen3 ASR
- Qwen3 ASR 1.7B - Speech to Text: Upload an audio file and Qwen3 ASR 1.7B transcribes it to text. Supports 52 languages, auto-detects the language, and handles noisy audio.
- Qwen3 ASR via TTS Audio Suite for SRT Builder: Upload your audio, get broadcast-ready captions back.
Platform Improvements/Bugfixes
- Serverless improvements.
- Websocket-based execution improves real-time responsiveness.
- Improved queue and island status tracking with new “downloading model” state.
- Subgraph ecosystem stability improvements with Nodes 2.0.
- Improved custom node discovery and detail pages.
- Fixed API node pricing badge inconsistencies.
- Updated Flex FloTime to 5 minutes and free API credits to $0.25.
- Added requested custom nodes, check full list of installed custom nodes on Floyo.
What’s New
- Generate Qwen-VL multimodal workflows directly in ComfyUI
- New Wan 2.7 video + image APIs now live (text → video, image → video)
- Nodes 2.0 + Subgraphs now supported → build reusable pipelines
- Serverless is now backing up Floyo infra to reduce queue wait time & improve scalability
New Workflows, you can try
- Wan 2.7
- Wan 2.7 - Text to Video: Generate video from a text prompt using Alibaba's Wan 2.7 model. Set your resolution, aspect ratio, and duration, then hit Run. Audio input supported.
- Multimodal AI (Qwen-VL)
- Qwen3-VL Image and Video Captioning: Get text descriptions of images and videos using Qwen3-VL, Alibaba's vision-language model.
- Comfyui-bitdance
- BitDance 14B - Text to Image: Generate photorealistic images from text prompts using BitDance 14B
- ComfyUI-Darkroom
- Darkroom Film Stock Emulation for Image Editing: Turn any digital image into a film photograph. Darkroom applies real film stock color science, grain patterns, halation, and print paper emulation to your images.
- DarkRoom Camera Raw Tools for Image Editing: Camera raw editing inside ComfyUI. White balance, exposure, HSL, clarity, vibrance, sharpening, noise reduction, and skin tone tools with before/after previews.
- Wan2.1
- Wan 2.1 InfiniteTalk: Wan 2.1 InfiniteTalk talking video from audio and a reference clip.
- Wan 2.1 Vid2Vid Style Transfer with Ditto: Upload any video, describe a new style, and Wan 2.1 rewrites every frame. Ditto keeps motion and structure intact across anime, Pixar, clay, and dozens more.
- Qwen Image Edit 2509 + Multi-Angle LoRA for Camera: Re-render your subject from any camera angle with Qwen Image Edit 2509 and a Multi-Angle LoRA. Pan, tilt, rotate, wide-angle, or close-up. No trigger word.
Platform Improvements/Bugfixes
- Serverless now live in production → faster startup times
- Subgraphs support → reusable workflow blocks (huge for power users)
- Fixed LoRA outputs not working in some workflows
- Publishing workflows now preserves correct model paths