Zhenzhen's AI Radar · Flux.3 Klein · 2026-11-15 Evening Report
This edition features 1 new versions: Strongly Recommended 0 · Recommended 0 · Worth Watching 1 · Included Only 0
Editor's Note This edition of Flux.3 Klein features only 1 new versions, listed in Column B, with no top recommendation: The_frizzy1's 12GB workflow v1.0.1 only rewrote the instructions while leaving the graph structure unchanged. The content consists of two graphs: BFL official int8 text-to-image (with optional bounding boxes) and editing with 1–4 reference images. BFL prompt expansion is enabled by default, and this version has no visible examples.
Worth Watching (1)
[B] FLUX 3 Image Klein 12GB — 5.6B int8 Text to Image + Edit with up to 4 Reference Images + BFL Prompt Upsampler · v1.0.1
Workflow · Version Update · Author The_frizzy1 · Published 11-15 20:54 · Downloads 1
Two workflows for running FLUX 3 Image Klein on a 12 GB GPU: text-to-image (with optional bounding boxes) and editing using 1–4 reference images. Based on BFL official int8 images, with BFL's built-in prompt expansion enabled by default, using only core nodes.
Illustration

Reasons for Recommendation
- Version update: v1.0.1 only rewrote the 'How to use' instructions in two files and moved the download instructions, correcting the peak VRAM for text-to-image to 11.2 GB. The author states the graph structure remains unchanged.
- Serves users with 12 GB GPUs who want to use Klein for text-to-image and multi-image reference editing; the author lists all required files, placement directories, and VRAM data on RTX 3060.
- The author explains why prompt expansion is enabled by default: Klein is trained on long scene descriptions, so short prompts produce mediocre results. The expansion model examines the input image first in edit mode.
How to Use
- ComfyUI 0.41 or higher; download flux3-image-klein-int8-convrot, qwen3vl_4b text encoder, prompt expansion model, and flux2-vae. Default is 1 MP.
- The 3th and 4th input images in the edit graph are muted by default; select them and press Ctrl+M to enable. Disable 'enable upsampler' only when writing your own long descriptions.
Risks and Limitations
- This version only changes instructions; users who have installed v1.0.0 do not need to update. There are no visible examples (2 hidden).
- The author states that 4 MP runs on 12 GB but causes banding in the sky, recommending staying at 1 MP. The author states it has not yet been actually run on ComfyUI 0.45.3.
- The FLUX Kommunity Licence restricts commercial use; read the license before selling.
- The card illustration is an AI-generated schematic, not the actual model output; the author has no publicly available standard-level examples.
Data: Civitai Public API (base model Flux.3 Klein), inclusion window 11-15 08:31 → 11-15 21:01 (Beijing Time). Reasons for recommendation have been reviewed by editors; download counts only reflect attention levels.