Private LLM - Lokaler KI-Chat

Name: Private LLM - Lokaler KI-Chat
Price: 5.99 EUR
Rating: 3.4 (10 reviews)
Author: Numen Technologies Limited

Offline privater KI-Chat

5,99 €

10 Bewertungen
Alter

13+
Jahre
Kategorie

Dienstprogramme
Entwicklungsteam

Numen Technologies Limited
Sprache

EN
Englisch
Größe

1,4
GB

Entdecken Sie den ultimativen KI-Assistenten für iOS mit Fokus auf Datenschutz: Private LLM Steigern Sie Ihre Produktivität und Kreativität auf iPhone und iPad mit Private LLM, dem führenden KI-Assistenten, der für Ihre Privatsphäre entwickelt wurde. Erhältlich als Einmalkauf ohne Abonnement, bietet die App fortschrittliche KI, die zu 100 % offline und vertraulich auf Ihrem Gerät arbeitet. Warum Private LLM Ihr idealer KI-Begleiter ist: * Auswahl an Top-KI-Modellen: Wählen Sie aus einer Vielzahl von Open-Source-LLM-Modellen (Llama 3.1, Llama 3.2, Gemma 2, Phi-3, Qwen 3, Mistral u.v.m.), die mit OmniQuant-Quantisierung für iOS optimiert sind. Passen Sie Ihr KI-Erlebnis für Brainstorming, Programmierung oder tägliche Fragen an. * Integration mit Siri & Kurzbefehlen: Verbessern Sie die KI-Interaktion mit Siri-Befehlen und anpassbaren Kurzbefehlen. Private LLM fügt sich nahtlos in Ihr Apple-Ökosystem ein. * Anpassbare Interaktionen: Gestalten Sie die KI-Antworten mit benutzerdefinierten System-Prompts nach Ihren Wünschen. * Absolute Privatsphäre: Ihre Gespräche bleiben auf Ihrem Gerät. Unsere vollständig offline arbeitende KI stellt sicher, dass keine Daten kompromittiert werden – keine Internetverbindung nötig. * Familienfreigabe & Offline-Nutzung: Ein Einmalkauf schließt die Familienfreigabe ein. Laden Sie Modelle herunter und nutzen Sie die App überall, auch ohne Internet. Eine vollständige Liste der unterstützten Modelle und deren Spezifikationen finden Sie unter privatellm.app/models. Private LLM ist mehr als nur ein Chatbot; es ist ein umfassender KI-Begleiter, der Ihre Privatsphäre respektiert und vielseitige Hilfe bietet. Ob Sie schreiben, programmieren oder Antworten suchen, Private LLM passt sich sicher an Ihre Bedürfnisse an. Steigern Sie noch heute Ihre Produktivität und kreativen Projekte mit dem privatesten KI-Assistenten für iOS. Private LLM übertrifft generische Wrapper-Apps (llama.cpp/MLX) in drei Punkten: 1. Schnellere, hochoptimierte mlc-llm Inferenz-Engine. 2. Modernste OmniQuant-Quantisierung für Modelle statt naiver Methoden. 3. 100 % native App (C++, Metal, Swift) mit tiefer iOS/iPadOS-Integration, im Gegensatz zu überladenen, nicht-nativen Apps (Electron/Flutter). Bitte beachten Sie: Private LLM unterstützt nur textbasierte LLMs. Die Modellunterstützung variiert je nach Geräteleistung.

Bewertungen & Rezensionen

3,4

von 5

10 Bewertungen

Great app, but one missing piece
4. Okt.

Spakie4
The app itself is great. There’s honestly nothing more I could ask for… except for one thing. That one thing, is chat history. And because of that missing, it really does ruin my experience a lot. If this does get added any time soon, I’d be very interested in trying it again. The UI could also get updated to the Liquid Glass from Apple, but that’s not too important. More importantly is chat history getting added. And adding an option to regenerate a response without having to modify it would also be greatly appreciated. Otherwise, great app!
Great app, but one missing piece
4. Okt.

Spakie4

The app itself is great. There’s honestly nothing more I could ask for… except for one thing. That one thing, is chat history. And because of that missing, it really does ruin my experience a lot. If this does get added any time soon, I’d be very interested in trying it again. The UI could also get updated to the Liquid Glass from Apple, but that’s not too important. More importantly is chat history getting added. And adding an option to regenerate a response without having to modify it would also be greatly appreciated. Otherwise, great app!
Very good fast and good/average accurate
17.12.2023

Markusxc
Works very well on iPad Pro 11‘‘ with M1 Chip. I would like to get support for more LLMs (currently only 2 are available).

Entwickler-Antwort
Thanks for the review! This is already supported in the macOS version of the app and we're currently working on adding support for multiple downloadable models in the iOS version of the app.
Very good fast and good/average accurate
17.12.2023

Markusxc

Works very well on iPad Pro 11‘‘ with M1 Chip. I would like to get support for more LLMs (currently only 2 are available).
Entwickler-Antwort 09.12.2023
Thanks for the review! This is already supported in the macOS version of the app and we're currently working on adding support for multiple downloadable models in the iOS version of the app.
Difficult
28. März

Blvkcwood
Is there huggingface support? Can i download/copy a model from there? I have no experience
Difficult
28. März

Blvkcwood

Is there huggingface support? Can i download/copy a model from there? I have no experience
Constant crashes
31. Jan.

Alexander Dorn
Unfortunately doesn’t work on the iPad 9th Gen with any other model than the useless demo model. If you have anything lower than a Silicon based iPad just a waste of money. On Silicon based Macs there are other much better free solutions.

Entwickler-Antwort
Thanks for the review. On-device LLM inference is memory intensive. Your 9th gen iPad that was released in 2021 and discontinued last year in 2024, only has 3GB of RAM, of which apps can only use about 2GB. This is sufficient to run only the smallest of models. Our app's USP is in high performance LLM inference and state of the art model quantization, which leads to higher accuracy compared to naive round to nearest quantization used by all the free "solutions". If you're willing to compromise on either of those, then this app likely isn't right for you. Please go to reportaproblem.apple.com and request a refund. Also, unlike the inferior "free solutions" that you're advocating for, this app runs on macOS, iOS and iPadOS. We're not aware of any other app, free or otherwise that's capable of doing this.
Constant crashes
31. Jan.

Alexander Dorn

Unfortunately doesn’t work on the iPad 9th Gen with any other model than the useless demo model. If you have anything lower than a Silicon based iPad just a waste of money. On Silicon based Macs there are other much better free solutions.
Entwickler-Antwort 2. Feb.
Thanks for the review. On-device LLM inference is memory intensive. Your 9th gen iPad that was released in 2021 and discontinued last year in 2024, only has 3GB of RAM, of which apps can only use about 2GB. This is sufficient to run only the smallest of models. Our app's USP is in high performance LLM inference and state of the art model quantization, which leads to higher accuracy compared to naive round to nearest quantization used by all the free "solutions". If you're willing to compromise on either of those, then this app likely isn't right for you. Please go to reportaproblem.apple.com and request a refund. Also, unlike the inferior "free solutions" that you're advocating for, this app runs on macOS, iOS and iPadOS. We're not aware of any other app, free or otherwise that's capable of doing this.