One prompt, one seed, many sampler and scheduler pairs. You get one labelled grid and a time table, fastest first.
Use it to find the fastest setting that still writes your words right.
Watch the video: https://www.youtube.com/watch?v=_RROWGtbN4g
What you need
- Stable Yogi Nodes (SY Nodes) in ComfyUI/custom_nodes (Sampler Sweep, Decode, Image Grid and Save Result)
- ComfyUI 0.43 or newer
- A MOSS picture model in models/diffusion_models. Ours are free here on CivitAI: Realism by Stable Yogi Moss (photo look) or Muse by Stable Yogi Moss (glamour). Both make a finished picture in 8 steps at CFG 1.
- qwen3.5_9b_w4a8_gptq.safetensors (7.3 GB) in models/text_encoders, type moss_image (Apache-2.0)
- flux2-vae_small.safetensors (250 MB) in models/vae
How to use it
- Drag the .json onto the ComfyUI canvas.
- Write your prompt; put the words you want on the picture in quotes.
- In Sampler Sweep SY, list one "sampler + scheduler" per line and keep the seed fixed.
- Press Run. The grid lands in ComfyUI/output and the time table shows every pair, fastest first.
- The first pair also pays for loading the model, so ignore its first time or run twice.
The step-by-step guide, the video and more free workflows are on my portal.

