LFM3-Omni-2B Runs Offline on iPhone at ~80 tok/s

A Reddit user reports running LFM3-Omni-2B fully offline on an iPhone 18 Pro Max at about 80 tokens per second using custom WebGPU kernels in a React Native app, highlighting on-device multimodal inference potential.

• Developer unveils CRAP fine-tune that emits x86 binaries2026-11-01
• Ternary Bonsai 2 models released, 2B version runs locally in browser2026-10-16
• Talus: 23M-parameter diffusion model for game terrain runs in browser2026-10-10
• Developer rebuilds Mozilla's Orbit with local inference2026-10-09
• Hugging Face open-sources fastest WebGPU kernels for local AI2026-09-30