On-device AI: Running LFM2.5-2.6B on iOS
helloiamleonie · x · 2026-08-31
Developer built an iOS app running the LiquidAI LFM2.5-2.6B model.
- Workflow: Fine-tuned with QLoRA for structured tool routing, converted to MLX, and quantized to 4-bit.
- Features: Triggers real iOS features (Reminders, Maps, timers) from natural language and tracks live vitals (tok/s, memory) with manual controls.
- Insight: Demonstrates the potential of on-device AI for system integration and monitoring.
More from coding & agent
- FrankenMarkdown: zero-dependency Rust engine renders Markdown to self-contained HTML/PDF via WASM — doodlestein · 2026-08-31
- Swyx breaks down pre-training vs post-training and what's holding back agents — iamrobotbear · 2026-08-31
- Using Codex with Blender beats hiring a 3D modeler for speed — natesiggard · 2026-08-31
- Exploring Object Detection Prompt Optimization for LLMs in DSPy — lateinteraction · 2026-08-31
- Report: OpenAI bought tens of thousands of Mac minis for agent training — ZeroStateReflex · 2026-08-31
- Meta^n: Recursive Self-Improvement Hits 0.331 on ARC-AGI-2 — burny_tech · 2026-08-31