Bonsai 27B brings 27-billion-parameter LLM inference to smartphones

xenova released Bonsai 27B, a quantized language model small enough to run inference directly on phones without cloud connectivity. For indie developers building offline-capable AI features or privacy-focused apps, having a capable model that fits in mobile RAM eliminates infrastructure costs and data transmission risks.