Liquid AI's LFM2.5-24B-A2B Enables Agent Tasks on Laptop Hardware

Liquid AI has released LFM2.5-24B-A2B, a purpose-built language model designed specifically for running agentic AI tasks on local hardware. The 24-billion parameter model achieves competitive performance against larger models while maintaining the memory footprint and inference speed needed for on-device deployment on standard laptops.

This release is particularly significant for practitioners building autonomous agents that require low latency and privacy guarantees. The model demonstrates superior performance to comparably-sized alternatives like Gemma 4 26B-A4B and Qwen 3.6 35B-A3B, making it an attractive choice for applications requiring tool calling, decision-making, and multi-step reasoning on consumer hardware.

The availability of LFM2.5-24B-A2B in GGUF format and optimised quantisations means developers can deploy agentic workflows immediately using existing inference engines like llama.cpp and Ollama, without requiring specialised infrastructure or cloud API dependencies.

Read the full article on Startup Fortune.

Source: Startup Fortune · Relevance: 9/10