| --- |
| title: FLUX.2-klein-4B GGUF |
| emoji: "\U0001F3A8" |
| colorFrom: purple |
| colorTo: indigo |
| sdk: gradio |
| sdk_version: 5.0.0 |
| app_file: app.py |
| pinned: false |
| --- |
| |
| # FLUX.2-klein-4B GGUF — CPU Inference |
|
|
| > ⚠️ **This model is gated.** You MUST complete steps 1-3 below or the Space will crash on startup. |
|
|
| ## Setup (Required) |
|
|
| ### Step 1: Accept Model License |
| Visit the model page and accept the license agreement: |
| **https://huggingface.co/unsloth/FLUX-2-klein-4B-GGUF** |
|
|
| ### Step 2: Create HF Token |
| Go to **https://huggingface.co/settings/tokens** → Create new token → **Read** permission |
|
|
| ### Step 3: Add Token to Space |
| Go to your Space → **Settings** → **Repository secrets** → Add: |
| - Name: `HF_TOKEN` |
| - Value: `hf_xxxxxxxxxxxxxxxxxx` |
|
|
| ## Hardware |
| - **No GPU**, CPU only |
| - 18 GB RAM |
|
|
| ## Performance Tips |
|
|
| | Resolution | Steps | Est. Time | |
| |-----------|-------|-----------| |
| | 256×256 | 8 | 1-3 min | |
| | 384×384 | 8-10 | 2-5 min | |
| | 512×512 | 8-12 | 4-10 min | |
| | 768×768 | 8-12 | 10-25 min | |
|
|
| First generation triggers PyTorch JIT compilation and is slower. Subsequent runs are faster. |
