Can AI Agents Really Run on Your Phone? 📱
Imagine having AI agents that can run entirely on your device, from laptops to phones, without relying on the cloud. Sounds too good to be true? 🤔
We've made it possible with LFM2.5-2.6B, a reliable agentic model that's small and fast enough for everyday hardware. This means developers can deploy agents everywhere, keep data private on the device, and scale usage without a cloud inference bill. 💸
What sets LFM2.5-2.6B apart? It's competitive with models 4x larger on tool use, instruction following, and multi-step agentic tasks, and it's been trained inside popular agentic harnesses to improve compatibility. Plus, it achieves efficient inference speeds of 220 tok/s on an Apple M5 Max and 113 tok/s on an AMD Ryzen CPU, in under 2.5 GB of memory. 🚀