Kolors
Diffusion model based on SDXL and ChatGLM3 text encoding, supporting both English and Chinese text rendering with realistic skin textures and lighting.
The brutal truth
Incredible portrait fidelity and bilingual typography. ChatGLM3 text encoder handles nuances that CLIP misses completely.
Files and Versions
Direct inspected tree artifacts from Hugging Face. Verified filenames, exact byte sizes, model weight shards, and configuration definitions.
Kwai-Kolors/KolorsFiles and versions are hosted on Hugging Face. You can view all branches, commits, model weights, and tokenizer artifacts directly.
Inspect Kwai-Kolors/Kolors /tree/main↗Decision matrix
✓ RUN / USE THIS IF…
- •You generate Asian subjects, fashion, or bilingual text
- •You have a 12GB or 16GB GPU
- •You want Apache 2.0
- •Your 24GB GPU comfortably fits the model weights (~10GB VRAM) with dedicated headroom for inference.
✕ WALK AWAY IF…
- •You want Western fantasy styles (SD3.5 Large is better)
- •Your GPU has less than 10GB VRAM and you cannot tolerate slow CPU/system RAM offload paging.
- •You prefer a plug-and-play web app interface and don't want to manage local model checkpoints, drivers, or runtimes.
- •You need specialized frontier reasoning that exceeds this model's parameter envelope.
True cost of ownership
About
Diffusion model based on SDXL and ChatGLM3 text encoding, supporting both English and Chinese text rendering with realistic skin textures and lighting.
What it does
Download the weights from Hugging Face, pick a quantisation that matches your GPU, and run it locally with Ollama, llama.cpp, vLLM, or a compatible runtime. Photorealistic portraits & bilingual text in images runs on your own hardware — no per-token bill, no third-party data handling.
Key features
Bilingual English and Chinese text in-image generation
Very natural human portrait and skin shading
Permissive Apache 2.0 license
Use cases
Explore visual directions before committing to production.
Place products in editorial scenes without a photo shoot.
Generate campaign imagery with consistent art direction.
Illustrate articles and decks with a consistent visual voice.
Quantization & Hardware Fit
VRAM Hardware Compatibility & Memory Allocation
| Quant | VRAM | Quality / Fit on 24GB | Install / Run Command (COMFYUI) | |
|---|---|---|---|---|
| FP8● ACTIVE | 10 GB | Standard · ✓ Fits (14GB spare) | comfy model download --url https://huggingface.co/Kwai-Kolors/Kolors --relative-path models/diffusion_models | |
| FP16 | 16 GB | Standard · ✓ Fits (8GB spare) | comfy model download --url https://huggingface.co/Kwai-Kolors/Kolors --relative-path models/diffusion_models |
Buyer doubts, cleared
Can I actually run Kolors on my 24GB GPU?
What's the fine print on the license?
Does the Text-to-image actually hold up in practice?
User reviews
We evaluated Kolors against two hosted alternatives. It won on cost-per-output but lost on latency. For batch work it is a clear yes.
The output reads naturally once you tune the prompt. It has cut my first-draft time roughly in half.
The FP8 build is the sweet spot. Anything larger needs serious hardware for marginal quality gains.
Kolors handled our batch jobs without surprises. The API is well documented and the rate limits are reasonable.
