LFM2.5-8B-A1B-oQ6

This model is an MLX oQ6 quantized version of LiquidAI/LFM2.5-8B-A1B, quantized using oQ (oMLX v0.3.12) mixed-precision quantization.

Quantization details

• Model type: lfm2_moe
• Bits: 6
• Group size: 64
• Format: MLX safetensors

Base Model: LFM2.5-8B-A1B

LFM2.5 is a new family of hybrid models designed for on-device deployment by Liquid AI. It builds on the LFM2 architecture with extended pre-training and reinforcement learning.

Model Details

Property	Value
Total parameters	8.3B
Active parameters	1.5B
Number of layers	24 (18 double-gated LIV conv + 6 GQA)
Training budget	38 trillion tokens
Context length	131,072
Vocabulary size	128,000
Languages	English, Arabic, Chinese, French, German, Japanese, Korean, Portuguese, Spanish

Recommended Generation Parameters

• temperature: 0.2
• top_p: 80
• repetition_penalty: 1.05

Chat Template

LFM2.5 uses a ChatML-like format:

<|startoftext|><|im_start|>system
You are a helpful assistant trained by Liquid AI.<|im_end|>
<|im_start|>user
What is C. elegans?<|im_end|>
<|im_start|>assistant

Citation

@article{liquidAI20268BA1B,
  author  = {Liquid AI},
  title   = {LFM2.5-8B-A1B: Personal Assistant On Your Laptop},
  journal = {Liquid AI Blog},
  year    = {2026},
  note    = {www.liquid.ai/blog/lfm2-5-8b-a1b},
}

@article{liquidai2025lfm2,
  title   = {LFM2 Technical Report},
  author  = {Liquid AI},
  journal = {arXiv preprint arXiv:2511.23404},
  year    = {2025}
}