LiquidAI has released LFM2.5-24B-A2B, a 24 billion parameter reasoning model that utilizes only 2 billion active parameters for efficient on-device operation. This model is designed for fast, private agents and demonstrates competitive performance against larger models on agentic and tool-use tasks. The release also includes a GGUF version, LiquidAI/LFM2.5-24B-A2B-GGUF, optimized for various inference engines like llama.cpp, vLLM, and Ollama, enabling broader accessibility and faster inference speeds. AI
IMPACT Enables faster, more private AI agents on-device by offering a high-performance model with reduced computational requirements.
RANK_REASON Model release from a recognized AI lab (LiquidAI) with detailed technical specifications and performance claims. [lever_c_demoted from frontier_release: ic=2 ai=1.0]
Read on Hugging Face Trending Models →
- Docker
- Google Colab
- Hugging Face
- Kaggle
- LiquidAI
- LiquidAI/LFM2.5-24B-A2B
- Qwen3.6-35B-A3B
- SGLang
- transformers
- vLLM
- LiquidAI/LFM2.5-24B-A2B-GGUF
- llama.cpp
- Ollama
AI-generated summary · Google Gemini · from 2 sources. How we write summaries →
