# Pocket LLMG — Personal Local Text Assistant

> Personal Local Text Assistant (iPhone/iPad app by 斌 肖.)

- Source: https://appshunter.io/ios/app/pocket-llmg/id6811747722 (this page in markdown: same URL + `.md`)
- Developer: [斌 肖](https://appshunter.io/developer/6805707752)
- Category: Utilities, Lifestyle
- Price: Free
- Age rating: 4+
- Requires: iOS 26.0 · 33 MB
- Languages: American English
- Released: 2026-09-22
- Data updated: 2026-10-29

## What is Pocket LLMG?

LocalGemma With Gemma E2B Text model support

MLX Optimized: High-performance inference using Apple’s MLX framework.

Developer-Grade Features:

Smart Tool Calling: Seamless, automated function loops by downloading Gemma E2B Text model only.
Memory Efficiency: Advanced KV-cache quantization and memory management to prevent crashes.
Jinja-Powered: Accurate prompt templating for complex models.
Privacy First: 100% local execution. No cloud. No leaks.
Experience bleeding-edge, local intelligence. Optimized for Apple Silicon.
Users need use Data(Wi-Fi or Cellular) for downloading Gemma E2B Text model from HF site only.

## Is Pocket LLMG safe?

Not enough ratings yet to assess Pocket LLMG.

## Is Pocket LLMG legit?

Pocket LLMG appears legitimate, with a few gaps. The AppsHunter Legitimacy Score for Pocket LLMG is 65 out of 100 points across 7 verification signals from App Store and developer data.


## Version history (last 5 releases)

### 1.2.0 — 2026-10-27

About now includes the full Gemma model license. It links to the Gemma Terms of Use
and the Gemma Prohibited Use Policy, shows the required notice, and notes that the
downloaded model is a modified 4-bit build. The app is not affiliated with Google.

Nothing else changes: your chats, settings and downloaded model stay as they are.

### 1.1.9 — 2026-10-26

The app is now called Pocket LLMG. It is the same app with the same model, and all
your saved chats, settings and downloaded model carry over.

Replies show up sooner: the first words appear immediately instead of waiting for the
next screen update, and long answers scroll smoothly while they are written. Typing
also stays responsive in long conversations.

### 1.1.8 — 2026-10-25

Your conversation now stays complete. Earlier versions quietly dropped everything
before the last exchange once a chat reached its third message — the full transcript
now stays on screen and in your saved history, while the model still reads only a
short recent window to stay within memory limits.

Turning Thinking or tool calling on or off no longer keeps a second copy of the model
in memory, which was the main reason the app could be closed mid-chat. A failed model
load can be retried, and a download can always be cancelled.

Long replies stay in view while they stream in, and the calculator no longer crashes
on very large results.

### 1.1.7 — 2026-10-20

move clear chat to main menu.

### 1.1.6 — 2026-10-17

Liquid effect support, use the short name of LocalGemma.

## More apps by 斌 肖

- [Pocket LLM Chat](https://appshunter.io/ios/app/pocket-llm-chat/id6805865627)
- [RunningAir](https://appshunter.io/ios/app/runningair/id6806567936)
- [Pure Air Run](https://appshunter.io/ios/app/pure-air-run/id6806569464)
- [Loop Habits](https://appshunter.io/ios/app/loop-habits/id6812269397)
- [Spendly Expense Track](https://appshunter.io/ios/app/spendly-expense-track/id6813476422)

All apps by 斌 肖: https://appshunter.io/developer/6805707752

---

*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 2026-10-29.*