Liquid AI releases QAD versions of LFM2.5 models for efficient edge deployment

Liquid AI has released a Quantization-Aware Distillation (QAD) version of its LFM2.5 series of small AI models. This new version is designed to reduce memory usage while minimizing performance degradation, making the models more suitable for low-spec devices like smartphones and laptops. The QAD technique involves distilling a high-accuracy teacher model to improve the performance of a quantized student model. Four LFM2.5 models (230M, 350M, 1.2B-Instruct, and 2.6B) have received this QAD treatment and are available as open models. AI

IMPACT Enables more efficient deployment of AI models on resource-constrained edge devices.

RANK_REASON Release of new quantized AI model versions with performance benchmarks. [lever_c_demoted from research: ic=1 ai=1.0]

Read on Mastodon — mastodon.social →

• Galaxy S26 Ultra
• Hugging Face
• LFM2.5
• LFM2.5-1.2B-Instruct
• LFM2.5-230M
• LFM2.5-2.6B
• LFM2.5-350M
• Liquid AI
• MacBook Pro
• NucBox EVO-X2
• Quantization-Aware Distillation
• Raspberry Pi 5

• model release
• product

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →