Try LFM • Docs • LEAP • Discord

LFM2.5-350M-MLX-6bit

MLX export of LFM2.5-350M for Apple Silicon inference.

LFM2.5-350M is a compact multilingual base model built on LiquidAI's hybrid architecture, combining convolutional and attention layers for efficient long-context processing.

Model Details

Property	Value
Parameters	350M
Precision	6-bit
Group Size	64
Size	296 MB
Context Length	128K

Use with mlx

pip install mlx-lm

from mlx_lm import load, generate
from mlx_lm.sample_utils import make_sampler

model, tokenizer = load("LiquidAI/LFM2.5-350M-MLX-6bit")

response = generate(
    model,
    tokenizer,
    prompt="The capital of France is",
    max_tokens=100,
    sampler=make_sampler(temp=0.7),
    verbose=True,
)

Other Precisions

• LFM2.5-350M-MLX-bf16 (676 MB)
• LFM2.5-350M-MLX-8bit (381 MB)
• LFM2.5-350M-MLX-6bit (296 MB)
• LFM2.5-350M-MLX-5bit (254 MB)
• LFM2.5-350M-MLX-4bit (212 MB)

License

This model is released under the LFM 1.0 License.