# LLM Farm — Run LLM > iOS/macOS app for running and customizing large language models. (iPhone/iPad app by Artem Savkin.) - Source: https://appshunter.io/ios/app/llm-farm/id6461209867 (this page in markdown: same URL + `.md`) - Developer: [Artem Savkin](https://appshunter.io/developer/1702380219) - Category: Developer Tools - Price: Free - Rating: 4.25/5 from 60 App Store ratings · 35 written reviews indexed - Age rating: 4+ - Requires: iOS 16.4 · 504 MB - Languages: English - Released: 2023-12-13 - Data updated: 2026-06-10 - Monetization: free - User reviews in markdown: https://appshunter.io/ios/app/llm-farm/id6461209867/reviews.md ## What is LLM Farm? LLMFarm is an iOS and MacOS app to work with large language models (LLM). It allows you to load different LLMs with certain parameters. # Features * Various inferences * Various sampling methods * Metal * Model setting templates # Inferences * LLaMA * GPTNeoX * Replit * GPT2 + Cerebras * Starcoder(Santacoder) * RWKV * Falcon * MPT * Bloom * StableLM-3b-4e1t * Qwen * Gemma * Phi * Mamba * Others # Multimodal * LLaVA 1.5 models * Obsidian * Bunny * MobileVLM 1.7B/3B models Note: For Falcon, Alpaca, GPT4All, Chinese LLaMA / Alpaca and Chinese LLaMA-2 / Alpaca-2, Vigogne (French), Vicuna, Koala, OpenBuddy (Multilingual), Pygmalion/Metharme, WizardLM, Baichuan 1 & 2 + derivations, Aquila 1 & 2, Mistral AI v0.1, Refact, Persimmon 8B, MPT, Bloom select llama inferece in model settings. Sampling methods * Temperature (temp, tok-k, top-p) * Locally Typical Sampling * Mirostat * Greedy * Grammar ## Key features - Load various LLMs - Multiple inference options - Diverse sampling methods - Metal GPU acceleration - Multimodal model support - Model setting templates ## Recent user reviews (10 of 35) All indexed reviews: https://appshunter.io/ios/app/llm-farm/id6461209867/reviews.md ### 1/5 — I can’t copy text from LLM Farm for external use of my research, useless ! *2025-05-04, version 1.4.3* Please correct this ASAP ! ### 5/5 — Great app *2025-04-30, version 1.4.3* Incredibly good on iphone16 pro max. ### 4/5 — Copying text *2025-04-21, version 1.4.3* Ability to copy text out of the chat window would be great. Otherwise a good app. Would love to see newer models for download. ### 5/5 — It works! *2025-04-08, version 1.4.3* Excellent app allowing easy setup for LLMs, looking forward to future updates to allow for attached photos or docs ### 4/5 — More models!!!!!! *2025-04-05, version 1.4.3* Stable Diffusion and can’t copy the llm’s answer ### 4/5 — Very neat but needs some QOL improvements. *2025-03-19* The fact that I can run an LLM on my iPad is insane. The fact that this is free is also insane. I imagine this app took a ton of work to create, so hat’s off. The one thing that I find irritating however is that you cannot edit responses (either the LLM’s or your own) and you can’t reroll a response, which is something that is pretty important especially with heavily quantized models that sometimes don’t “get it” the first time around. Also an indicator of the current context size would be nice. There’s also some controls that are unexplained like what looks like a refresh button that turns in to a checkmark. I have no idea what that does. Keep up the great work though! ### 5/5 — Great, but… *2025-03-12* Overall an amazing app. I really support allowing for anybody to grab models off huggingface and use them locally, but there are still a few issues: 1. Low RAM devices. I suspect certain models are being masked based on memory, but I dont see any newer models. Ive made and run some mamba/deepseek/qwen 3B param models locally with 4GB, so I know its possible. 2. audio models (speech to text, text to speech, and audio to audio) models are possible to run at low sample rates. The only other app I use for local language models is specifically because of this. 3. This is more long term, but I would like to see support for loading regular pytorch models (yaml/json for architecture, .pt/.pth for weights). All in all, this is a great app, even without any changes. Thank you!! ### 3/5 — Good lLM app but *2025-02-26* I tried the app is great app, on image it show that user can import images to the chat, when I tried it and tap on button all app become grey without open the Image Photo Album!, I think it’s a bug on current version I hope it fixed, other than that, it’s the best Local LLM app so far ! ### 4/5 — Works well but copy failed *2025-02-18* I would give 5 stars except copy function doesn’t work. ### 1/5 — What is this *2025-02-15, version 1.4.3* Selected the smallest model and it still won’t work or reply. Also you to stay on the download screen and watch download or repeat ## Frequently asked questions about LLM Farm ### What is LLM Farm? LLM Farm is an application designed for iOS and macOS that allows users to load and run various large language models (LLMs). It provides a platform for experimenting with different models and their parameters, including advanced inference and sampling techniques. ### What kind of LLMs can I run with LLM Farm? The app supports a wide array of LLM architectures, including LLaMA, GPTNeoX, GPT2, Starcoder, RWKV, Falcon, MPT, Bloom, Qwen, Gemma, Phi, Mamba, and many others. It also offers support for multimodal models like LLaVA. ### Does LLM Farm support multimodal models? Yes, LLM Farm supports multimodal models such as LLaVA 1.5, Obsidian, Bunny, and MobileVLM. This allows users to work with models that can process both text and image inputs. ### What are the sampling methods available? The application offers several sampling methods to control LLM output, including Temperature (temp, tok-k, top-p), Locally Typical Sampling, Mirostat, Greedy, and Grammar sampling. These methods allow for fine-tuning the creativity and coherence of generated text. ### Is LLM Farm free to use? Based on the provided information, LLM Farm is available for free and does not have in-app purchases or advertisements. This allows users to explore its features without any cost. ## Version history (last 5 releases) ### 1.4.3 — 2025-01-31 ## Changes * llama.cpp updated to b4562 * Added support MiniCPM-1B, Minicpm-omni, Deepseek-R1-Qwen distill, PhiMoE, DeepSeek V3 models * Added supprot Minerva 7B, Deepseek MoE v1 & GigaChat models, Qwen2VL, Falcon3 models * Added support InfiniAI Megrez 3b, OLMo, QRWKV6, Llama-3_1-Nemotron models * Ask question to PDF document content from shortcut, see shortcut example in documentation. * Max RAG answer count option * Metal improvements * Fixed duplication of documents in the index * Fixed keyboard overlapping of settings items * Some metal fixes and improvements * Some other fixes ### 1.4.0 — 2024-11-12 Changes: * llama.cpp updated to b3982 * Added RAG support for pdf documents * Chat settings UI improvements * Added text summarization shortcut * Some metal improvemets * Added support for Chameleon * Fixed a bug causing the message copy dialog to be missing on some devices. * Fixed some other errors ### 1.3.9 — 2024-10-15 Changes: * llama.cpp updated to b3837 * Added support for llama 3.2, 3.1, RWKV, MiniCPM(2.5, 2.6, 3), Chameleon models * Added Metal support for Mamba * Added Llama 3.2 download links (working with LLaMa3 Instruct template) * Added Phi 3.5 download links (working with Phi 3 template) * Added Bunny template and download link * Some metal improvements * Fixed some errors for gemma2, DeepSeek-V2, ChatGLM4, llama 3.1, T5, TriLMs, BitNet models * Fixed some other errors ### 1.3.4 — 2024-10-07 * Added Gemma2, T5, JAIS, Bitnet support, GLM(3,4), Mistral Nemo * Added OpenELM support (270M only so far) * Added the ability to change the styling of chat messages. * Added a built-in demo chat for new users. (The ChatML template is used) * Some detokenizer fixes - LoRa and FineTune are temporarily disabled due to the need for code refactoring due to a large number of errors. ### 1.3.0 — 2024-06-28 Changes: * LLaMA.cpp updated to b3190 * Added support for DeepseekV2, GPTNeoX (Pythia and others) * Added support for Markdown formatting * Added support for using history in Shortcuts * Added Flash Attention support * Added NPredict option * Save/Load context state. Now it is possible to continue the dialog with the model even after the program is reopened. * Chat settings such as sampling are now applied without reloading the chat. * Skip tokens option. Allows you to specify tokens that will not be displayed in the results. May be useful for phi3 and qwen. * Metal and CPU inference improvements * Sampling and eval improvements * Some fixes for phi-3 and MiniCPM * Fixed some errors * Added Qwen template --- *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 2026-06-10.*