# Local LLM: MITHRIL — Local AI Models On-Device

> On-device AI language model runner for mobile devices. (iPhone/iPad app by DEPLOY FORWARD LLC.)

- Source: https://appshunter.io/ios/app/local-llm-mithril/id6751945393 (this page in markdown: same URL + `.md`)
- Developer: [DEPLOY FORWARD LLC](https://appshunter.io/developer/1834702526)
- Category: Productivity, Developer Tools
- Price: Free
- Rating: 4.40/5 from 13 App Store ratings · 7 written reviews indexed
- Age rating: 12+
- Requires: iOS 15.1 · 52 MB
- Languages: American English
- Released: 2025-09-08
- Data updated: 2026-07-25
- Monetization: free
- User reviews in markdown: https://appshunter.io/ios/app/local-llm-mithril/id6751945393/reviews.md

## What is Local LLM?

Run quantized large language models directly on your iPhone. No cloud, no internet required.

Access state-of-the-art quantized AI models optimized for mobile hardware. Download GGUF-format models that compress billion-parameter networks into mobile-friendly sizes while maintaining performance.

COMPLETE MODEL SUITE
• Llama 3.2 1B/3B (Meta) - Q4/Q8 quantization
• Gemma 3 270M/2B/9B (Google) - IQ4_NL optimization  
• Qwen 2.5 0.5B-7B (Alibaba) - Multiple quantization levels
• LLaVA 1.5/1.6 (Vision) - Multimodal image understanding
• Direct integration with Hugging Face model repository

TECHNICAL FEATURES
• GGML/llama.cpp inference engine
• Metal GPU acceleration on Apple Silicon
• Dynamic context window management (2K-8K tokens)
• Retrieval-Augmented Generation (RAG) with embeddings
• Real-time streaming with token/second metrics
• SQLite conversation storage with vector search

SYSTEM REQUIREMENTS
Models run efficiently when file size ≤ available RAM. Recommended minimum 6GB RAM for larger models. iPhone 15 Pro/Pro Max optimal. iOS26 for Apple foundation model.

Zero telemetry. Zero data transmission. Pure local AI computing.

## Key features

- Run AI models locally on iPhone
- No cloud or internet required
- Access quantized AI models
- Supports GGUF-format models
- Metal GPU acceleration
- RAG with embeddings
- SQLite conversation storage

## Recent user reviews (7 of 7)

All indexed reviews: https://appshunter.io/ios/app/local-llm-mithril/id6751945393/reviews.md

### 4/5 — Good but

*2026-02-20*

Needs a bit more stuff, like allowing to edit the system prompt. The rog feature is really good.

### 5/5 — bug in iPhone xs

*2026-01-15*

Bug Report: Model Persistence and Startup Errors (iPhone XS)
App Version: Mithril (Local LLM) Device: iPhone XS iOS Version: [ ios 18.3.2]

Issue Summary

I am experiencing two main issues regarding model management and app startup:

Redundant Download Prompt: After successfully importing and downloading a custom model, the app prompts me to download it again every time I select it.

ExecuTorch Error: Every time the app opens, an "ExecuTorch error" appears, even though the model is already present on the device.

Steps to Reproduce

Import a custom model and complete the download process.

Select the model for use.

Observe the prompt asking to download the model again. (Current workaround: I close/force-quit the app to bypass this).

Re-open the app and observe the "ExecuTorch error" notification upon startup.

Expected Behavior

The app should recognize that the model is already stored locally and load it immediately without re-triggering the download prompt.

The app should launch without ExecuTorch errors if the model files are valid and present.

Additional Notes

Despite the error messages, the model does eventually work if I bypass the prompts.

I am using an iPhone XS, so there may be specific optimizations or memory constraints related to older A12 Bionic hardware that are triggering these flags.

### 4/5 — Works as described so far

*2026-01-03*

I’ve just started using the app and it’s working as described. There is one critical feature missing. It must have a way to quickly copy the generated text. Each request and response needs to be able to be copied easily. Preferably with a button or a shortcut. I will update my rating to 5 stars a fast as possible when this feature is added. If there is a stretch goal for this app I would like to be able to use text2image models as well. Great job, and keep up the great work.

### 5/5 — Best app for local LLMs

*2026-01-03*

Currently the best App Store iOS app for running local LLMs. Despite its many shortcomings it still gets 5 stars simply because there is currently no competition to this app to my knowledge. It is the only app capable of running any model that’s compatible with llama.cpp. This allows the use of custom fine tuned models downloaded from Hugging Face which is essential as 4B parameters is about the maximum that an iPhone can currently handle. It really needs some improvements such as the ability to copy prompts and answers to the system clipboard. Currently the best workaround is to take a screenshot and use the iOS Live Text feature to copy the text to clipboard although this is impractical for long answers. 
Some other issues include the export data feature. The exported .json file is terribly formatted. 

The app has limited support for markdown and formatting.  Failing to render tables for example. 

Thinking models also have their output truncated making it impossible to see the models entire process. 

Browsing conversation history is annoying. It lacks search feature and is limited to past 100 conversations which is exasperated by the apps poor quality .json exports. 

A previous app I have used called “Locally AI” has a very good UX, but has tons of technical limitations.

### 5/5 — Best app for local LLMs

*2025-12-27*

Currently the best App Store iOS app for running local LLMs. Despite its many shortcomings it still gets 5 stars simply because there is currently no competition to this app to my knowledge. It is the only app capable of running any model that’s compatible with llama.cpp. This allows the use of custom fine tuned models downloaded from Hugging Face which is essential as 4B parameters is about the maximum that an iPhone can currently handle. It really needs some improvements such as the ability to copy prompts and answers to the system clipboard. Currently the best workaround for anyone interested is to export data (at the bottom of the screen) and parse the export file, Or take a screenshot and use the iOS Live Text feature to copy the text to clipboard although this is impractical for long answers.

### 5/5 — Perfect for local models

*2025-12-15*

This works better then I would expect on a phone. Would love to be able to host an openai style service on it with it.

### 1/5 — Clunky, impossible to paste API

*2025-11-25*

The app is clunky with no instructions. It doesn't allow API key to be pasted. Impossible to contact developer.

## Frequently asked questions about Local LLM

### What AI models can I run with MITHRIL?

MITHRIL supports a complete model suite including Llama 3.2, Gemma 3, Qwen 2.5, and LLaVA 1.5/1.6. It also integrates directly with the Hugging Face model repository for a wide selection of GGUF-format models.

### Does MITHRIL require an internet connection?

No, MITHRIL is designed to run AI models entirely on your device. There is no cloud or internet connection required for its core functionality, ensuring zero data transmission.

### What are the system requirements for MITHRIL?

Models run efficiently when their file size is less than or equal to your available RAM. A minimum of 6GB RAM is recommended for larger models, with the iPhone 15 Pro/Pro Max being optimal. iOS 26 is required for Apple foundation models.

### How does MITHRIL handle conversations?

MITHRIL uses SQLite for conversation storage, which includes vector search capabilities. This allows for efficient retrieval and management of your chat history locally.

### What is the update frequency for MITHRIL?

The latest version, 1.6.2, was released on October 13, 2025. The app is actively developed, with a release date of September 8, 2025, suggesting regular updates.

### Is MITHRIL suitable for younger users?

MITHRIL has an age rating of 12+, indicating it is intended for teenagers and older. This is due to the nature of the AI models and their potential outputs.

### Does MITHRIL have ads?

No, MITHRIL is ad-free. The app is completely free to download and use without any advertisements interrupting your experience.

### What is the performance like on different Apple devices?

MITHRIL utilizes Metal GPU acceleration on Apple Silicon for optimal performance. While it can run on various iPhones, iPads, and iPods, the iPhone 15 Pro/Pro Max is recommended for the best experience with larger models.

## Version history (last 5 releases)

### 1.6.2 — 2025-10-13

-fixed ability to voice chat with Apple Foundation Model
-smoothed onboarding for voice chat- now if no models are downloaded yet a modal will pop up and prompt you to download one of the 3 whisper models instead of automatically downloading them all
-if content block is hit with foundation model in voice chat, a modal will popup and explain that Apple limits content then give option to start new chat or switch to open source model

### 1.6.1 — 2025-10-08

new model added:


  - Name: LiquidAI LFM2 8B (Q4)
  - Download URL: https://huggingface.co/LiquidAI/LFM2-8B-A1B-GGUF/resolve/main/LFM2-8B-A1B-Q4_0.gguf
  - Size: ~4.7 GB
  - Parameters: 8B

ONLY WORKS ON HIGH END DEVICES - 6-8GB RAM AND GREATER PLEASE DO NOT RUN ON OLDER IPHONES !

### 1.6.0 — 2025-10-03

-onboarding recognizes iOS and gives option to skip straight into chat if on iOS26 with foundation model
-some small bugs in the foundation model chat - file RAG chat now works with foundation model on file uploads
-removed LLAVA model as users reported errors and bugs specifically with that model

### 1.5.0 — 2025-10-01

-added voice chat to chat directly with LLMs
-voice audio detection so you don't need to manually start and stop your recording, just continuous conversation flow
-voice chat will be developed further into the future integrating new voices and more

### 1.4.1 — 2025-09-28

-fixed model selection menu bug

## Apps similar to Local LLM

| App | Rating | Price | Category |
| --- | --- | --- | --- |
| [focus timer & tracker - foku.](https://appshunter.io/ios/app/focus-timer-and-tracker-foku/id6749248718) | 5.0 (1) | Free | Productivity |
| [MobileLLM](https://appshunter.io/ios/app/mobilellm/id6743194941) | — | Free | Productivity |
| [RePrompt: AI Prompt Builder](https://appshunter.io/ios/app/reprompt-ai-prompt-builder/id6747730380) | — | Free | Productivity |
| [Control - LLM Interface](https://appshunter.io/ios/app/control-llm-interface/id6752229004) | — | Free | Productivity |
| [NoemaAI](https://appshunter.io/ios/app/noemaai/id6751169935) | 4.2 (16) | Free | Productivity |
| [LLM Pigeon](https://appshunter.io/ios/app/llm-pigeon/id6746935952) | 3.8 (4) | Free | Productivity |

## More apps by DEPLOY FORWARD LLC

- [infipedia](https://appshunter.io/ios/app/infipedia/id6752109334)
- [FPV Drone Operator Manual](https://appshunter.io/ios/app/fpv-drone-operator-manual/id6752651844)

All apps by DEPLOY FORWARD LLC: https://appshunter.io/developer/1834702526

## Related topics

[local llm vision](https://appshunter.io/ios/topics/local-llm-vision) · [local llm](https://appshunter.io/ios/topics/local-llm) · [llm hub](https://appshunter.io/ios/topics/llm-hub) · [nllm](https://appshunter.io/ios/topics/nllm) · [gldm](https://appshunter.io/ios/topics/gldm) · [local llm ai app](https://appshunter.io/ios/topics/local-llm-ai-app) · [llvm](https://appshunter.io/ios/topics/llvm)

---

*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 2026-07-25.*