Liquid AI has released LFM2.5-230M, its smallest model to date, designed for on-device inference on edge hardware like phones and robots. This 230-million-parameter model excels at data extraction and tool use, outperforming larger models on specific benchmarks like IFEval and IFBench. While not suited for complex reasoning tasks such as math or coding, its small footprint and efficient architecture allow for fast inference, making it ideal for local data processing and lightweight agentic workloads. AI
IMPACT Enables more capable AI agents and data processing directly on edge devices, reducing reliance on cloud infrastructure.
RANK_REASON Release of a small, specialized model focused on on-device inference and specific tasks, rather than a frontier model release.
- Gemma 3 1B
- Hugging Face
- LFM2.5-230M
- Liquid AI
- llama.cpp
- MLX
- NVIDIA Jetson Orin
- ONNX
- Qwen3.5-0.8B
- Raspberry Pi 5
- Galaxy S25 Ultra
- SGLang
- Unitree G1
- vLLM
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →
