Maple-Preview – Ternary 20B MoE running at 120 tok/s on a iPhone

In plain words

Maple-Preview is a large language model that runs directly on iPhones, delivering inference speeds of 120 tokens per second. Built on a ternary 20 billion parameter mixture-of-experts architecture, it enables users to run advanced AI locally without relying on cloud services. The software is designed for iPhone users seeking fast, on-device language processing capabilities.

written from the facts on this page · September 2026

• 
  Ollama v0.19Apr 2026 · ollama.com · ▲413

  Massive local model speedup on Apple Silicon with MLX

• 
  Run an 80B Qwen in 4.3 GB of RAM on a Mac, and a 35B on an iPhoneAug 2026 · github.com · ▲312

  Swiftlet is a Swift and Metal runtime that runs large Qwen Mixture-of-Experts models locally on Apple devices by streaming expert weights from storage, enabling 35B and 80B models to run with low RAM, including on iPhone. - leonickson1/Swiftlet

• 
  Gan.AI TTS Model & API Playground2024 · ▲344

  First TTS model to support all 22 Indic languages + English

• 
  MiMo-V2-FlashDec 2025 · mimo.xiaomi.com · ▲241

  Ultra-fast 309B MoE model for coding & agents

• 
  Qwen3.6-35B-A3BApr 2026 · qwen.ai · ▲124

  The open sparse MoE model for agentic coding

• 
  Lightning V3Apr 2026 · smallest.ai · ▲323

  Text-to-Speech built for Voice Agents

• TL
  The load-bearing vocabulary of Claude▲703

  Life & fun · 11d ago · louisabraham.github.io

• 
  Articos▲390

  Launch with confidence, not gut instinct

  Life & fun · 12d ago · articos.com

• 
  Nex▲350

  Claude Cowork for high-volume GTM workflows

  Life & fun · 4d ago · nex.ai

• 
  iPhone app takes simultaneous images from 2 lenses, fuses into 1 photo▲329

  Photosynthesis fires two of your iPhone

  Life & fun · 29d ago · photosynthesis.camera

• CC
  Creatium Coach▲319

  Your multimedia mentor that takes you from mid to great

  Life & fun · 11d ago · producthunt.creatium.info

• 
  SoloUno▲310

  Take control of hair pulling, nail biting & skin picking

  Life & fun · 29d ago · solouno.io

• TL
  The load-bearing vocabulary of Claude▲703

  Life & fun · 11d ago · louisabraham.github.io

• 
  Hey Noah▲641

  A proactive AI executive assistant for founders

  AI · Aug 2026 · heynoah.io

• 
  Clipto MCP▲640

  Let agents source clips from terabytes of your local video

  Work · 19d ago · clipto.com

• 
  Simple algorithm and color space to generate diverse skin tones▲627

  Hello HN! I found that picking out plausible but diverse skin tones for my digital art and game development projects was kind of difficult, and I got curious about if there was a way to define a color space that made it easy. I've built a color picker and procedural generation algorithm based on the space as well as a bunch of other fun js features and demos throughout the page that use the equations. If you find it interesting, I have lots of explanations of how I built it and what properties the space has. The methodology might be a bit shaky, but hopefully the result is as helpful for…

  Life & fun · Aug 2026 · toneyalexander.github.io

• 
  AdAnt AI▲608

  Claude for viral, high-converting social ads

  AI · Aug 2026 · adant.ai

• 
  I trained a 125M model to autocomplete piano on-device▲598

  I trained a 125M-parameter transformer to autocomplete piano performances in real time (~108 notes/sec on an iPhone 15). The idea is basically GitHub Copilot or Tabnine, except instead of prompting it with code, you prompt it by playing a few notes on a MIDI piano. The model then continues what you played, entirely on-device. The app is free if anyone wants to try it. Happy to answer questions about the model, training, Core ML, or the many things that didn't work.

  AI · 18d ago · simedw.com