ComfyUI error: Blurry/noisy output

Schnell is a 4-step distilled model (source). CFG 1.0 is stated by the source for the Flux checkpoint versions.

Flux.1 Schnell 4-step text-to-image

Fast 4-step distilled Flux text-to-image.

Estimated VRAM

~11.89 GB — ⚠️ ESTIMATE (unverified real-world usage). Estimate, not a measured figure: weight-size floor = flux1-schnell.safetensors (11.89 GB, sizes verified on Hugging Face/GitHub); halved for fp8 weight_dtype per ComfyUI_examples Flux page. Activations add more; ComfyUI offloading can reduce it. Source, checked 2026-10-13.

Models required

flux1-schnell.safetensors — 23.783 GB ✅ verified 2026-10-13 (listing) → models/unet
t5xxl_fp8_e4m3fn.safetensors — 4.894 GB ✅ verified 2026-10-13 (listing) · sha256 7d330da4816157540d6bb7838bf63a0f02f573fc48ca4d8de34bb0cbfd514f09 → models/text_encoders
clip_l.safetensors — 0.246 GB ✅ verified 2026-10-13 (listing) · sha256 660c6f5b1abae9dc498ac2d21e1347d2abdb0cf6c0c0c8576cd796491d9a6cdd → models/text_encoders

Tested GPUs

No verified test reports yet.

Nodes

UNETLoader DualCLIPLoader VAELoader KSampler VAEDecode

Common errors & fixes

Blurry/noisy output: Schnell is a 4-step distilled model (source). CFG 1.0 is stated by the source for the Flux checkpoint versions. ✅ per source

Low-VRAM tips

⚠️ General guidance, not individually verified.

fp8 weight_dtype
fp8 T5

Source & credit

Workflow by comfyanonymous (ComfyUI_examples). Get it from the original: https://comfyanonymous.github.io/ComfyUI_examples/flux/. License: Flux.1 Schnell: Apache 2.0.