Patreon
Home / Videos / Make AI Images 2x Faster on RTX 50 GPUs: NVFP4 With CUDA 13

Make AI Images 2x Faster on RTX 50 GPUs: NVFP4 With CUDA 13 in ComfyUI, Quality Tested

SwarmUI full tutorial from a fresh Windows install: our installers, model downloader with 517+ models and 86 presets turn your PC into a complete AI studio.

Published Jan 11, 2026 · 11,061 views · Watch on YouTube

Get the app and files

SwarmUI Full Tutorial: 1-Click Install, 4 MP Images, HD Video & Lip Sync

SwarmUI Full Tutorial: 1-Click Install, 4 MP Images, HD Video & Lip Sync

SwarmUI full tutorial from a fresh Windows install: our installers, model downloader with 517+ models and 86 presets turn your PC into a complete AI studio.

ComfyUI Latest Updates: New Models, Presets and Improvements

ComfyUI Latest Updates: New Models, Presets and Improvements

ComfyUI update log with the latest four releases; V144 adds ten presets for Fibo 2.0, MOSS Image Edit and FastH3 Omni, for 106 ready presets in total.

ComfyUI Install & FLUX 3 Klein: The Complete Beginner Tutorial

ComfyUI Install & FLUX 3 Klein: The Complete Beginner Tutorial

ComfyUI beginner tutorial: install on Windows and learn FLUX 3 Klein with 16 core examples, precise image edits, RTX 5090 tests and launcher tuning.

SimplePod AI: Cheap Cloud GPU Alternative to RunPod and Vast AI

SimplePod AI: Cheap Cloud GPU Alternative to RunPod and Vast AI

SimplePod AI is a cheap cloud GPU service with a ready SECourses template, so all our RunPod scripts work unchanged, with permanent disks faster than RunPod.

BF16 vs GGUF Q8 vs FP8 Scaled vs NVFP4: Quality and Speed Comparison

BF16 vs GGUF Q8 vs FP8 Scaled vs NVFP4: Quality and Speed Comparison

BF16, GGUF Q8, FP8 Scaled and NVFP4 precisions compared for speed and quality in ComfyUI with CUDA 13; NVFP4 ran 87% faster than GGUF Q8 on Z Image Turbo.

NVFP4 Model Converter Gradio App for FLUX: Windows & Linux

NVFP4 Model Converter Gradio App for FLUX: Windows & Linux

NVFP4 Model Converter app quantizes FLUX models to NVFP4, which runs 100%+ faster than BF16, FP8 or GGUF; the FLUX SRPO NVFP4 result is only 6.32 GB.

Chapters

  1. 0:00 New ComfyUI installer (CUDA 13, Torch 2.9.1, Triton + attention libs)
  2. 0:19 NVFP4 speedup claims vs real tests; why CUDA 13 enables new models
  3. 0:34 Prebuilt FlashAttention/SageAttention/xFormers for many GPUs (Windows + Linux)
  4. 1:00 Quality roadmap: FLUX2 Dev, Z Image Turbo, FLUX Dev (BF16/FP8/GGUF/NVFP4)
  5. 1:23 Downloader adds NVFP4: FLUX2 Dev, FLUX Dev (Context/Dev), Z Image Turbo
  6. 1:51 SimplePod AI intro: RunPod-style pods, cheaper rates, permanent storage
  7. 2:36 Musubi Tuner FP8 Scaled: quality myths vs GGUF + why scaled matters
  8. 3:10 Quantization & precision (FP32/BF16/FP8/GGUF) + Qwen3 low-VRAM encoders
  9. 3:34 ComfyUI v73 zip: CUDA 13 included; update NVIDIA drivers only (v72 deprecated)
  10. 4:13 Update steps: overwrite zip, delete venv, run install/update .bat
  11. 5:02 Python: 3.10 recommended (supports 3.10-3.13); fresh vs update
  12. 6:02 New installer flow: uv speed, standalone use, backend libs detected
  13. 7:12 Stability flags: --cache-none vs --disable-smart-memory (OOM/stuck fixes)
  14. 7:54 SwarmUI presets: 32 presets supported; drag/drop + auto model downloader
  15. 8:25 Update SwarmUI model-downloader zip (extract + overwrite)
  16. 8:49 Download bundles/models (Z Image Turbo Core + NVFP4 options)
  17. 9:25 Update/launch SwarmUI; point to updated ComfyUI backend + set args
  18. 10:32 Live gen test: Z Image Turbo BF16 @1536x1536
  19. 11:29 Switch to NVFP4: VRAM cache behavior; 1024x1024
  20. 12:36 FLUX2 Dev quality: FP8 Scaled vs NVFP4 side-by-side comparisons
  21. 13:33 Speed chart: FLUX2 NVFP4 about 193% faster than FP8 Scaled
  22. 14:10 Z Image Turbo quality: BF16 vs NVFP4 vs FP8 Scaled (quant method)
  23. 15:25 FLUX Dev: FP8 Scaled approx GGUF Q8; NVFP4 currently shows degradation
  24. 16:45 What precision means + model size examples (FP32/BF16/FP8 Scaled/NVFP4)
  25. 18:07 Practical recommendations: BF16 best; avoid FP16; raw FP8 vs FP8 Scaled
  26. 19:43 GGUF explained: block quant, slower runtime; use only when RAM is too low
  27. 21:36 Precision hierarchy recap + when to pick FP8 mixed/scaled over GGUF
  28. 21:58 SimplePod setup: register, add credits, open template link
  29. 22:31 Template config + RunPod price comparison (disk, ports, GPU selection)
  30. 24:02 Persistent volume: create + mount to /workspace
  31. 25:11 Launch RTX Pro 6000 pod; SimplePod vs RunPod pricing differences
  32. 26:29 Temp vs persistent disk: deleting instance wipes temp data - backup!
  33. 26:55 JupyterLab: upload zips, apt install zip, unzip ComfyUI in workspace
  34. 27:48 Run install script; unzip SwarmUI; start the model downloader
  35. 29:02 Downloader path for ComfyUI + folder structure; download Z Image Turbo bundle
  36. 30:08 Start ComfyUI; confirm CUDA 13 + Torch 2.9.1; connect via port 3000 Direct
  37. 31:08 Preset demo: Z Image Turbo Quality 1; fix VAE path; monitor VRAM
  38. 33:18 File Browser Direct: download outputs/models fast; upload files back
  39. 34:41 Restart server; install/start SwarmUI; open Cloudflared URL
  40. 36:26 SwarmUI backend: /workspace/ComfyUI/main.py + args; import presets
  41. 37:27 Download FLUX2 Core + NVFP4; share model paths between SwarmUI & ComfyUI
  42. 39:27 FLUX2 NVFP4 generation @2048x2048; VRAM usage + step speed
  43. 40:43 Cloud GPU pitfall: diagnosing a power-capped GPU
  44. 41:28 Resume: re-run template w/ volume; reconnect fast
  45. 45:02 Wrap-up: SimplePod pros (direct/secure, cheaper storage)

About this video

Finally NVFP4 models has arrived to ComfyUI thus SwarmUI with CUDA 13. NVFP4 models are literally 100%+ faster with minimal impact on quality. I have done grid quality comparison to show you the difference on FLUX 2, Z Image Turbo and FLUX 1 of NVFP4 versions. To make CUDA 13 work, I have compiled Flash Attention, Sage Attention & xFormers for both Windows and Linux with all of the CUDA archs to support literally all GPUs starting from GTX 1650 series, RTX 2000, 3000, 4000, 5000 series and more.

In this full tutorial, I will show you how to upgrade your ComfyUI and thus SwarmUI to use latest CUDA 13 with latest libraries and Torch 2.9.1. Moreover, our compiled libraries such as Sage Attention works with all models on all GPUs without generating black images or videos such as Qwen Image or Wan 2.2 models. Hopefully LTX 2 presets and tutorial coming soon too. Finally, I introduce a new private cloud GPU platform called as SimplePod like RunPod. This platform has all the features of RunPod same way but much faster and cheaper.

📂 Resources & Links:

ComfyUI Installers: [ https://www.patreon.com/posts/ComfyUI-Installers-105023709 ]

SimplePod: [ https://simplepod.ai/ref?user=secourses ]

SwarmUI Installer, Model Auto Downloader and Presets: [ https://www.patreon.com/posts/SwarmUI-Install-Download-Models-Presets-114517862 ]

How to Use SwarmUI Presets & Workflows in ComfyUI + Custom Model Paths Setup for ComfyUI & SwarmUI Tutorial: [ https://youtu.be/EqFilBM3i7s ]

SECourses Discord Channel for 7/24 Support: [ https://discord.com/invite/software-engineering-courses-secourses-772774097734074388 ]

NVIDIA NVFP4 Blog Post More: [ https://developer.nvidia.com/blog/introducing-nvfp4-for-efficient-and-accurate-low-precision-inference/ ]

⏱️ Video Chapters:

0:00 New ComfyUI installer (CUDA 13, Torch 2.9.1, Triton + attention libs)

0:19 NVFP4 speedup claims vs real tests; why CUDA 13 enables new models

0:34 Prebuilt FlashAttention/SageAttention/xFormers for many GPUs (Windows + Linux)

1:00 Quality roadmap: FLUX2 Dev, Z Image Turbo, FLUX Dev (BF16/FP8/GGUF/NVFP4)

1:23 Downloader adds NVFP4: FLUX2 Dev, FLUX Dev (Context/Dev), Z Image Turbo

1:51 SimplePod AI intro: RunPod-style pods, cheaper rates, permanent storage

2:36 Musubi Tuner FP8 Scaled: quality myths vs GGUF + why scaled matters

3:10 Quantization & precision (FP32/BF16/FP8/GGUF) + Qwen3 low-VRAM encoders

3:34 ComfyUI v73 zip: CUDA 13 included; update NVIDIA drivers only (v72 deprecated)

4:13 Update steps: overwrite zip, delete venv, run install/update .bat

5:02 Python: 3.10 recommended (supports 3.10-3.13); fresh vs update

6:02 New installer flow: uv speed, standalone use, backend libs detected

7:12 Stability flags: --cache-none vs --disable-smart-memory (OOM/stuck fixes)

7:54 SwarmUI presets: 32 presets supported; drag/drop + auto model downloader

8:25 Update SwarmUI model-downloader zip (extract + overwrite)

8:49 Download bundles/models (Z Image Turbo Core + NVFP4 options)

9:25 Update/launch SwarmUI; point to updated ComfyUI backend + set args

10:32 Live gen test: Z Image Turbo BF16 @1536x1536

11:29 Switch to NVFP4: VRAM cache behavior; 1024x1024

12:36 FLUX2 Dev quality: FP8 Scaled vs NVFP4 side-by-side comparisons

13:33 Speed chart: FLUX2 NVFP4 about 193% faster than FP8 Scaled

14:10 Z Image Turbo quality: BF16 vs NVFP4 vs FP8 Scaled (quant method)

15:25 FLUX Dev: FP8 Scaled approx GGUF Q8; NVFP4 currently shows degradation

16:45 What precision means + model size examples (FP32/BF16/FP8 Scaled/NVFP4)

18:07 Practical recommendations: BF16 best; avoid FP16; raw FP8 vs FP8 Scaled

19:43 GGUF explained: block quant, slower runtime; use only when RAM is too low

21:36 Precision hierarchy recap + when to pick FP8 mixed/scaled over GGUF

21:58 SimplePod setup: register, add credits, open template link

22:31 Template config + RunPod price comparison (disk, ports, GPU selection)

24:02 Persistent volume: create + mount to /workspace