FLUX.1 [schnell]
12B rectified-flow transformer, timestep-distilled to 1-4 sampling steps, released under Apache-2.0. Same architecture as FLUX.1 [dev] but trades a bit of fidelity for ~10x faster sampling and an unrestricted commercial license.
The pragmatic choice when you actually need to ship. Lose ~10-15% quality vs dev, gain a real commercial license and 10x faster inference. If you're building a product on top of FLUX, this is the one.
Overview
12B rectified-flow transformer, timestep-distilled to 1-4 sampling steps, released under Apache-2.0. Same architecture as FLUX.1 [dev] but trades a bit of fidelity for ~10x faster sampling and an unrestricted commercial license.
Strengths
- Apache-2.0 — only frontier-class open image model with no commercial restrictions
- 1-4 step sampling (~1-2s/image on RTX 4090) vs 20-50 for FLUX.1 [dev]
- Same 12B MMDiT backbone — keeps most of dev's prompt adherence and typography
- Drop-in replacement in diffusers / ComfyUI / FP8 / NF4 / GGUF tooling
Weaknesses
- Distillation visibly degrades fine detail and hands vs FLUX.1 [dev]
- Does not respond to CFG/guidance scale — locked to a single distilled trajectory
- Still needs ~24GB VRAM at BF16; consumer cards require quantization
- LoRA / fine-tuning ecosystem is thinner than dev's
Quantization variants
Each quantization trades model quality for file size and VRAM. Q4_K_M is the most popular starting point.
| Quantization | File size | VRAM required |
|---|---|---|
| Q4_K_M | 6.6 GB | 9 GB |
Get the model
HuggingFace
Original weights
Source repository — direct quantization required.
Hardware that runs this
Cards with enough VRAM for at least one quantization of FLUX.1 [schnell].
Models worth comparing
Same parameter band, plus what's one tier above and below — so you can decide what actually fits your hardware.
Frequently asked
What's the minimum VRAM to run FLUX.1 [schnell]?
Can I use FLUX.1 [schnell] commercially?
What's the context length of FLUX.1 [schnell]?
Source: huggingface.co/black-forest-labs/FLUX.1-schnell
Reviewed by RunLocalAI Editorial. See our editorial policy for how we research and verify model claims.
Related — keep moving
Verify FLUX.1 [schnell] runs on your specific hardware before committing money.