American startup Liquid AI presented a large language model LFM2-8B-A1B, which was created specifically to work on smartphones and laptops.
It has 8.3 billion parameters, but only 1.5 billion are active during query processing. Thanks to this, ensuring high performance and low power consumption.
To run the model on the iPhone 17 Pro, you will need the LocallyAIApp app and the model itself, which can be downloaded from Hugging Face. It weighs 9.34 GB. After loading the model, you need to import it into the application and select the context window options:
▶︎ 4 thousand tokens (~10 thousand characters): 9.5–10 GB RAM.
▶︎ 8 thousand tokens (~20 thousand characters): 10.5–11.8 GB RAM.
▶︎ 16 thousand tokens (~40 thousand characters): 12–14 GB RAM.
Given that the iPhone 17 Pro and 17 Pro Max are equipped with 12 GB of RAM, the maximum size of the context window that can be selected is 8 thousand tokens.
LFM2-8B-A1B puts a lot of stress on the smartphone, because while it is running, the graphics and RAM are turned on at maximum. However, the model still works quickly.
In terms of performance, the model approaches its competitors by 3–4 billion parameters and in some benchmarks it surpasses Lama-3.2-3B-Instruct and Qwen3-4B-Instruct-2507.
Source: Iphones RU
I am a professional journalist and content creator with extensive experience writing for news websites. I currently work as an author at Gadget Onus, where I specialize in covering hot news topics. My written pieces have been published on some of the biggest media outlets around the world, including The Guardian and BBC News.
