L
LiquidAI: LFM2-24B-A2B
◆ Strong7.5
/ 10
LFM2-24B-A2B is the largest model in the LFM2 family of hybrid architectures designed for efficient on-device deployment. Built as a 24B parameter Mixture-of-Experts model with only 2B active parameters per...
Specifications
| Attribute | Value |
|---|---|
| Lab | Liquid |
| Tags | Fast Open Weight |
| Overall Score | 7.5/10 |
| Release Date | 2026-02 |
| Context Window | 128,000 tokens |
| Input Price / 1M | $0.03 |
| Output Price / 1M | $0.12 |
| Input Modalities | Text |
| Output Modalities | Text |
Strengths
- Hybrid architecture with 24B total / 2B active params
- Designed for efficient on-device deployment
- Very low pricing at $0.03/$0.12
- MoE for scalable inference
Weaknesses
- Very limited capabilities vs larger models
- Narrow use case for on-device/edge
- Small context window for its class
Best For
- On-device AI applications
- Ultra-low-cost inference
- Edge deployment scenarios
Sources & Further Reading
Related Models
Scores are aggregated from public benchmarks (MMLU, HumanEval, MATH, GSM8K, LMSYS) and normalized to a 1–10 scale. Methodology →