rag vs ChatOKF
Price, ratings, monetisation and update history for both apps, side by side — with what reviewers say about each.
rag
Engage in private AI conversations with locally-running language models. Upload documents like PDFs and DOCX to query them with AI, and visualize your document embeddings in a 3D atlas. Choose from various AI providers and models optimized for your device's performance.
- 100% offline AI chat
- Document RAG with PDF/DOCX
- Multiple AI model providers
- 3D embedding visualization
- Real-time device monitoring
- Conversation history with search
Read full descriptionHide full description
RAG Bot — Private AI on Your Device Have natural conversations with locally-running large language models. Your chats stay on your device with full conversation history and search. Your Documents, AI-Powered Upload PDFs or DOCX files. RAG Bot generates embeddings so you can chat with your documents using retrieval-augmented generation. Visualize your document embeddings in an interactive 3D atlas. Choose Your Model Browse and download from three AI providers (Cactus, ExecuTorch, llama.rn). Filter by type (LLM or embedding), size, capabilities like vision and tool calling — find the right model for your hardware. Built for Performance Real-time device monitoring shows memory and storage usage so you always know what your device can handle. Optimized for Apple Silicon with multiple inference backends. Key Features 100% offline — no internet required after model download Multiple model providers and quantization options Document RAG with PDF, DOCX support Conversation history with full search Vision and tool-calling capable models Real-time device resource monitoring 3D embedding visualization
ChatOKF
Read full descriptionHide full description
A capable AI assistant that lives on your iPhone and iPad. The model runs on your device's own chip, so nothing you type is uploaded and nothing is logged. Once a model is downloaded, ChatOKF works in airplane mode. No account. No sign-in. No subscription. No ads, no analytics, no telemetry. WHAT YOU CAN DO - Ask anything: check the math on a loan, plan a week of dinners, untangle a confusing letter. - Show it a picture: a photo of a menu, a street sign or a whiteboard, straight from your library. Some models also listen to audio and watch video, frame by frame. - Hand it a document: a PDF, a Word file or a spreadsheet. It reads the tables inside and answers from the file, and one tap opens the page to check it. - Watch it think: the stronger models reason before they answer. Read the reasoning, tap Quick Answer to cut it short, or switch thinking off for speed. - Let it look things up: Wikipedia, today's headlines, the weather and web search are tools it can use, each behind its own switch. - Keep going: it suggests a sensible next question; swipe right to ask it. - Keep what it gives you: answers are typeset, tables and mathematics included. Search and rename your chats, share one as a PDF, and bring a deleted chat back within thirty days. - Talk instead of typing: tap the microphone and speak, and have replies read aloud in one of ten voices made on your device. MEMORY YOU CAN READ Tell it to remember something and it writes a short note, and the chat says what it wrote. Later questions are checked against your notes first, whichever model is answering. Every note is listed under Memories, to read or to delete. One switch, Total Recall, turns it all off. WHERE YOUR WORDS GO Nowhere, unless it looks something up. With Wikipedia on, your question is matched to an article on your device and only that article is downloaded. With web access on, the model's search terms go to a search engine and the pages it chooses are fetched. Settings, Privacy says exactly what leaves your device for each one. PICK YOUR MODEL Several open models are offered, and the list shows only what your device can hold, so every one you see will run. - Gemma 4 E2B and E4B: they see pictures, hear sound and watch video. - Qwen3.5 2B and 4B: good with tools and longer answers. - Ternary Bonsai 1.7B: small and fast, for devices with less memory. GOOD TO KNOW - The first launch downloads a model, from about half a gigabyte to a few gigabytes. Use Wi-Fi; after that, no network is needed. - Devices with more memory are offered stronger models, and newer chips answer faster. - Answers are generated and can be wrong. Check anything that matters. ChatOKF is written in Swift on Metal, and its source is on GitHub. Models are provided by their authors under their own licenses, including the Gemma Terms of Use.
Screenshots
Verdict
rag and ChatOKF are too close to call on the numbers: across 5 compared parameters — price, ads, in-app purchases and device support among them — neither pulls far enough ahead to decide it. What separates them is feature set and interface, not measurable difference.
Scored on Price · Rating · Positive reviews · Number of ratings · Update frequency · Ad data · In-app purchases · Monetization · Best chart rank · Devices · Requires iOS
Both are free to download. Neither carries in-app purchases, so what you see is what you pay.
| Parameter | rag | ChatOKF |
|---|---|---|
| Price | Free | Free |
| Rating | 1.0 (1 ratings) — better | — |
| Positive reviews | 0.0% of reviews | — |
| Number of ratings | 1 — better | — |
| Update frequency | — | Fortnightly |
| Ad data | No | No |
| In-app purchases | No | No |
| Monetization | Free | — |
| Devices | iPhone, iPad | iPhone, iPad, Mac — better |
| Requires iOS | 17.0 — better | 18.0 |
| Further details — not scored | ||
| Size | 79 MB | 14 MB |
| Age rating | 4+ | 17+ |
| Developer | Kyle Essenmacher | Dmitry Kuznetsov |
In-app purchases
rag
No in-app purchases
ChatOKF
No in-app purchases









