Chroma
Stableby Lodestones
Community-distilled 8.9B model based on FLUX.1-schnell architecture. Apache 2.0 licensed alternative to Flux with competitive quality. Available in HD and Flash variants for different quality/speed tradeoffs.
VRAM requirements, GPU fit, and setup notes for Chroma, including 8GB/12GB fit guidance where relevant. Recommended runtimes: ComfyUI and Diffusers support. Best download size: ~8.9 GB at FP8.
- 8.9B DiT — based on FLUX.1-schnell
- Apache 2.0 — fully open for commercial use
- HD and Flash variants available
- Fast 4-step generation
Your hardware
Detecting...
Image Quality Benchmarks
Measured quality metrics for Chroma outputs.
How often humans prefer this model's output (0-100%)
Visual quality and composition rating (5-9 scale)
VRAM Requirements by Resolution and Precision
Compare which GPUs can run Chroma at different precisions. FP8 uses less memory than FP16 when available, and the grade shows how comfortably each GPU handles the workload.
FP16 (full precision)
| Resolution | VRAM Required | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 | 21.7 GB | A● | F● | F● | F● |
| 768×768 | 21.8 GB | A● | F● | F● | F● |
| 1024×1024 | 22.0 GB | A● | F● | F● | F● |
FP8 (~40% less VRAM)
| Resolution | VRAM Required | RTX 4090 24GB | RTX 3060 12GB | RTX 4060 8GB | MacBook Pro M4 Pro 24GB |
|---|---|---|---|---|---|
| 512×512 | 15.3 GB | S | D | F | A |
| 768×768 | 15.5 GB | S | D | F | A |
| 1024×1024 | 15.7 GB | S | D | F | A |
Run with Python
from diffusers import FluxPipeline
import torch
pipe = FluxPipeline.from_pretrained(
"lodestones/Chroma",
torch_dtype=torch.float16
)
pipe.to("cuda")
image = pipe(
prompt="your prompt here",
num_inference_steps=4,
height=1024,
width=1024,
).images[0]
image.save("output.png")Get started
Setup instructions for running Chroma locally
1. Download the model
Get the checkpoint from HuggingFace
2. Place in:
ComfyUI/models/checkpoints/3. Launch ComfyUI
python main.pyMemory Breakdown
VRAM allocation at 1024×1024 on RTX 4090 24GB (24 GB)
Estimated Generation Time
Time per image at 1024×1024, 28 steps, FP16.
Sample Outputs
Available Formats, Downloads & Setup
Download Chroma in the precision that matches your GPU. Lower precision usually means less VRAM pressure, while higher precision keeps more quality.
| Format | Präzision | Größe | Anbieter | |
|---|---|---|---|---|
| safetensorsEmpfohlen | FP16 | 17.8 GB | official | Herunterladen |
| safetensors | FP8 | 8.9 GB | official | Herunterladen |
LoRA Ecosystem
LimitedEarly LoRA ecosystem. Some Flux LoRAs may be partially compatible.
Related Workflows
You might also like
Frequently asked questions
FAQ — Chroma VRAM, Runtimes & Fit
How much VRAM does Chroma need?
Chroma (8.9B parameters) requires approximately 22.0 GB of VRAM at FP16 precision for standard 1024×1024 image generation. If you want a lighter setup, lower precisions like FP8 can reduce memory use when available.
Can I run Chroma on an 8GB GPU?
Chroma usually needs more than 8GB for comfortable local use. Check the VRAM table above for the exact resolution and precision trade-off.
Does Chroma work in ComfyUI and Diffusers?
Chroma is marked for ComfyUI and Diffusers support in our catalog, so those are the runtimes we recommend first for local setup. If your workflow uses another front end, check the model's available formats and workflow notes above before downloading.
Can I run Chroma on RTX 4090?
Yes, the RTX 4090 (24 GB VRAM) can run Chroma comfortably at FP16. Expected generation time is around ~7.3s per image at 1024×1024.
Does Chroma support ControlNet?
There are currently no known ControlNet adapters for Chroma. Check Hugging Face and Civitai for community-contributed adapters.
Does Chroma have LoRA support?
Early LoRA ecosystem. Some Flux LoRAs may be partially compatible. The LoRA ecosystem for Chroma is rated as "minimal". Each LoRA adds roughly 0.2 GB of extra VRAM.
How fast is Chroma?
On a reference GPU (RTX 4090 24GB), Chroma generates a 1024×1024 image in approximately ~7.3s at FP16 with 28 inference steps. Faster GPUs with higher memory bandwidth will produce images more quickly.
About Chroma
See also