4-bit Open on Hugging Face
Parameters
6.2B
Weights (est.)
≈ 3.5 GB
Mac memory (est.)
8 GB+
Size is estimated from the parameter count and quantisation. Memory allows a fifth extra for context plus 2 GB for the system; long contexts need more.
Run it
With mlx-vlm on an Apple silicon Mac:
pip install mlx-vlmpython -m mlx_vlm.generate --model mlx-community/Lance-3B-AWQ-INT4 --image photo.jpg --prompt "Describe this image"
Details
- Base model
- bytedance-research/Lance (quantized)
- Vendor
- bytedance-research
- Task
- Vision + text
- Quantisation
- 4-bit
- Licence
- apache-2.0
- Downloads
- 25 in the last 30 days · 282 all time · 0 likes
- Published
- 2026-05-26 · updated 2026-05-26 (5mo ago)
- Tags
- lance-mlx, qwen2_5_vl, lance, image-understanding, quantized, apple-silicon
Other MLX builds of bytedance-research/Lance
| Build | Quantisation | Weights | Mac memory | Downloads |
|---|---|---|---|---|
| Lance-3B-AWQ-INT4 (this one) | 4-bit | ≈ 3.5 GB | 8 GB+ | 25 |
| Lance-3B-8bit | 8-bit | ≈ 6.6 GB | 16 GB+ | 41 |
| Lance-3B-bf16 | bf16 | ≈ 12.4 GB | 16 GB+ | 49 |
| Lance-3B-Video-bf16 | bf16 | ≈ 12.9 GB | 24 GB+ | 766 |
| Wan2.2-VAE-Lance-bf16 | bf16 | — | — | 0 |