bytedance-research · First MLX build 21 May 2026
Lance
MLX
Image apache-2.0
A 6.4B parameter image to image model from bytedance-research, with 5 MLX builds, from 4-bit to bf16.
- Parameters
- 6.4B From Hugging Face
- Quantisation
- bf16 Lance-3B-Video-bf16
- Memory needed
- 17 GB Estimate, unified memory
- Download
- 12.9 GB Weights, estimated
- Downloads
- 776 This build, last 30 days
- Task
- Image to image Image
- MLX builds
- 5 On mlx-community
- Devices
- Mac Apple silicon
Memory
Will it fit?Needs about 17 GB at bf16
8 GB
16 GB
24 GB
32 GB
64 GB
128 GB
Fits with headroomTight, close other appsWon’t fit
Estimated from the size of the weights, plus a fifth for context and working memory and 2 GB for macOS and your apps. Long contexts need more.
Run it
Terminal · Hugging Face CLI$ pip install -U huggingface_hub
$ hf download mlx-community/Lance-3B-Video-bf16MLX builds
5 on mlx-community| Lance-3B-AWQ-INT4 | 4-bit | 3.5 GB | 6 GB | 25 |
|---|---|---|---|---|
| Lance-3B-8bit | 8-bit | 6.6 GB | 10 GB | 41 |
| Lance-3B-bf16 | bf16 | 12.4 GB | 16 GB | 50 |
| Lance-3B-Video-bf16 | bf16 | 12.9 GB | 17 GB | 776 |
| Wan2.2-VAE-Lance-bf16 | bf16 | — | — | 0 |
Sizes are estimated from Hugging Face’s parameter counts and the quantisation. Downloads are for the last 30 days.