# LLMVoice

> iOS 26 speech and on-device AI showcase using MLX Swift. (iPhone/iPad app by Andre Frelicot.)

- Source: https://appshunter.io/ios/app/llmvoice/id6753925010 (this page in markdown: same URL + `.md`)
- Developer: [Andre Frelicot](https://appshunter.io/developer/1055019120)
- Category: Developer Tools, Utilities
- Price: Free
- Rating: 5.00/5 from 1 App Store ratings
- Age rating: 4+
- Requires: iOS 26.0 · 18 MB
- Languages: American English
- Released: 2025-10-13
- Data updated: 2026-08-12
- Monetization: free
- User reviews in markdown: https://appshunter.io/ios/app/llmvoice/id6753925010/reviews.md

## What is LLMVoice?

LLMVoice — iOS 26 Speech & On-Device AI Showcase

A technical showcase of Apple's latest iOS 26 technologies for speech recognition and on-device AI, including MLX Swift for running open-source language models locally on Apple Silicon.

Built for Developers & Early Adopters

LLMVoice is made for developers and AI enthusiasts who want to explore Apple's newest frameworks and on-device AI capabilities, end to end, on a real device.

What It Does

• Real-time voice transcription using iOS 26 speech APIs
• AI-powered summarization with multiple model options
• Generated code shown with syntax highlighting
• Live HTML preview for generated content, with built-in diagnostics
• Optional "thinking" view: see the model's reasoning, or hide it for a clean output
• Modern SwiftUI interface with smooth streaming updates
• 100% on-device processing for complete privacy

Key Technologies

MLX Swift Framework
Apple's MLX (Machine Learning Extensions) enables efficient inference of quantized language models on Apple Silicon with optimized GPU acceleration.

Apple's On-Device Foundation Models
On supported devices (iPhone 15 Pro or later), use Apple's built-in system models for summarization — no extra downloads required.

Open-Weights Models via MLX
Choose from multiple open-source models running locally:
• Qwen2.5 (0.5B) — 150MB, 29+ languages, 32k context
• Gemma 3 (1B) — 300MB, 140+ languages, 32k context
• Llama 3.2 (1B) — 500MB, 8 languages (EN/ES/FR/DE/IT/PT/HI/TH)

All models use 4-bit quantization for mobile efficiency. Token limits are tuned per model for reliable results.

SpeechAnalyzer & SpeechTranscriber (iOS 26)
Low-latency, real-time on-device speech recognition using Apple's newest APIs.

Modern Swift & SwiftUI
Built with Swift 6, the Observable macro, async/await, and actor-based concurrency.

How It Works

1. Select an AI model (Apple's on-device models, or download an MLX model)
2. Tap the microphone to start recording
3. Speak naturally — real-time transcription appears
4. Tap stop when finished
5. The model generates a concise summary instantly
6. Summaries are saved locally and persist between sessions
7. Swipe to delete, tap to view the original transcription

Privacy-First

Everything runs on-device:
• Audio never leaves your device
• Local speech recognition and AI inference
• No cloud processing or internet required (except model downloads)
• No tracking, no data collection

Requirements

• iOS 26.0+ (iPadOS 26.0+ for iPad)
• Apple Silicon recommended
• Storage for models (150MB–500MB each)
• Microphone and speech recognition permissions

Good to Know

• Models are downloaded on first use from Hugging Face
• Apple's on-device models are available immediately on compatible devices, without downloads
• Primarily optimized for English; other languages vary by model

Perfect For

• Developers exploring iOS 26 and MLX Swift
• AI enthusiasts interested in on-device LLMs
• Students learning SwiftUI and modern concurrency
• Anyone curious about running open-weights models on mobile

Under the Hood

Demonstrates MLX Swift integration with Apple frameworks, including model downloading, GPU memory management, streaming inference with real-time metrics, and efficient on-device transformer execution.

Mac Catalyst Compatible

Runs on macOS 26+ with Apple Silicon for a unified cross-platform experience.

## Key features

- Real-time voice transcription
- AI-powered summarization
- Generated code with syntax highlighting
- Live HTML preview
- On-device processing
- MLX Swift framework integration
- Modern SwiftUI interface


## Frequently asked questions about LLMVoice

### What is LLMVoice?

LLMVoice is a technical showcase for iOS 26 speech recognition and on-device AI. It utilizes the MLX Swift framework to run open-source language models locally on Apple Silicon, offering features like real-time voice transcription and AI-powered summarization.

### What are the system requirements for LLMVoice?

LLMVoice requires iOS 26.0+ (or iPadOS 26.0+ for iPad). Apple Silicon is recommended for optimal performance. You will also need microphone and speech recognition permissions, and storage space for models ranging from 150MB to 500MB each.

### Does LLMVoice require an internet connection?

LLMVoice is designed for 100% on-device processing, meaning audio never leaves your device and AI inference is local. An internet connection is only required for downloading AI models from Hugging Face; otherwise, it functions offline.

### What AI models can I use with LLMVoice?

You can use Apple's built-in on-device foundation models for summarization on supported devices (iPhone 15 Pro or later). Additionally, LLMVoice supports several open-source models via MLX Swift, including Qwen2.5 (0.5B), Gemma 3 (1B), and Llama 3.2 (1B), all using 4-bit quantization for efficiency.

### How does LLMVoice handle privacy?

LLMVoice prioritizes privacy by performing all processing on-device. Audio data is not collected or sent to the cloud, and there is no tracking or data collection involved, ensuring complete user privacy.

### Is LLMVoice suitable for beginners?

While LLMVoice is built with modern Swift and SwiftUI, making it a great learning tool for developers, its primary audience is developers and early adopters exploring advanced iOS 26 and on-device AI capabilities. Beginners might find the technical nature challenging without prior Swift or AI knowledge.

### How often is LLMVoice updated?

LLMVoice was last updated on June 16, 2026, with version 1.1. Given its nature as a showcase for the latest iOS technologies, updates are likely to align with new framework releases and improvements from Apple.

### Does LLMVoice have ads?

No, LLMVoice does not contain advertisements. It is a free application focused on showcasing developer technologies without any ad-supported elements.

## Version history (last 2 releases)

### 1.1 — 2026-06-16

• Code blocks now feature syntax highlighting for easier reading
  • New toggle to show or hide the model's thinking process
  • Smoother streaming with reduced scroll jitter
  • Redesigned conversation list with quicker prompt actions
  • Smarter on-device models: per-model token limits and improved prompt handling for more reliable responses
  • More stable recording — no more accidental duplicate recordings
  • Various performance improvements and bug fixes

### 1.0 — 2025-10-13

No release notes.

## Apps similar to LLMVoice

| App | Rating | Price | Category |
| --- | --- | --- | --- |
| [Pacachat - Local AI Chat](https://appshunter.io/ios/app/pacachat-local-ai-chat/id6753087483) | 5.0 (1) | Free | Developer Tools |
| [SF Search](https://appshunter.io/ios/app/sf-search/id6738096244) | 5.0 (5) | Free | Developer Tools |
| [Symbolero](https://appshunter.io/ios/app/symbolero/id6744256360) | 5.0 (3) | Free | Developer Tools |
| [Python Code Develop](https://appshunter.io/ios/app/python-code-develop/id6742342498) | 5.0 (2) | $1.99 | Developer Tools |
| [HTTP Client: Geturl](https://appshunter.io/ios/app/http-client-geturl/id6753350349) | — | Free | Developer Tools |
| [App Localizer - Translator](https://appshunter.io/ios/app/app-localizer-translator/id6753215812) | — | Free | Developer Tools |

## More apps by Andre Frelicot

- [DecibelPeak](https://appshunter.io/ios/app/decibelpeak/id6752702602)
- [BLE UScope](https://appshunter.io/ios/app/ble-uscope/id6753627148)

All apps by Andre Frelicot: https://appshunter.io/developer/1055019120

## Related topics

[andre frelicot](https://appshunter.io/ios/topics/andre-frelicot)

---

*Data collected daily from the US App Store and indexed by [AppsHunter](https://appshunter.io/). User reviews are verbatim App Store reviews. Ratings, prices and chart positions refresh continuously; this snapshot is from 2026-08-12.*