Forem: FrederickSalazar

How to Install Local LLMs en your machine with OllamaFX 0.5.0

FrederickSalazar — Mon, 09 Mar 2026 14:58:01 +0000

Since launching OllamaFX, my goal has always been simple: create the lightest, fastest, and most useful native desktop client for Ollama. We wanted to build something that respects your privacy and doesn't rely on browser engines.

Today, I’m thrilled to announce our biggest leap forward. OllamaFX 0.5.0 is finally here. We are moving from a simple "local chatbot" to a true personal intelligence hub.

🚀 The Headline Features: More Power, Locally

1. RAG (Chat with Documents) – Total Privacy

The most requested feature is here. Retrieval-Augmented Generation (RAG).
OllamaFX can now index your local PDF and text files. When you chat, the model automatically pulls relevant context from your documents to answer your questions. This is perfect for analyzing massive technical manuals, private notes, or research papers—all processed entirely on your machine. No data ever leaves your computer.

2. Full Multimodal Vision Support

Stop typing descriptions and start analyzing reality. We have implemented native support for vision models like Llava and Bakllava. Simply drag and drop an image (diagrams, photos, whiteboard sketches) directly into the chat. Ask your local model to explain, debug, or extract data from the image. The potential for technical support and OCR is huge.

3. Native Auto-Updates

Staying up to date shouldn't be work. OllamaFX now includes a built-in update mechanism. The system will detect new releases, allow you to view the changelog, and update the application directly from the interface, ensuring you always have the latest features and stability fixes without manual downloads.

📂 The Developer Workflow: Getting Organized

A powerful AI tool is useless if you can't manage your output. As developers, we create dozens of chats daily. 0.5.0 addresses the clutter.

4. Categorized Chat Folders

You can now create custom folders in the sidebar. Move conversations into folders (e.g., 'Project Alpha', 'Research', 'Personal') to keep your workspace pristine. The drag-and-drop movement is fluid and instant.

5. The Recycle Bin (Undo Delete)

Accidentally deleting a complex prompt chain is devastating. 0.5.0 introduces a Recycle Bin. Deleted chats are moved to this section, where you can restore them or choose to empty the bin and delete them permanently. Peace of mind, unlocked.

🛠️ Performance, Stability, and the Full Changelog

Beyond the features, we spent significant time refining the application's core and the AtlantaFX (JavaFX) interface.

The Full 0.5.0 Changelog:

Feature: RAG (Retrieval-Augmented Generation) with local document indexing.

Feature: Complete native Multimodal Vision model support (Llava/Bakllava).

Feature: Integrated Auto-update system.

Workflow: Chat Folders for sidebar organization.

Workflow: Recycle Bin functionality with full restore capabilities.

UI: Fluid Drag-and-drop movement of chats between folders.

UI: AtlantaFX theme refinements (optimized Dark/Light modes).

UI: Enhanced Markdown rendering for code blocks.

Core: Detailed model info view (file size, format, quantization).

Stability: Optimized memory management for streaming responses.

Fix: Resolved intermittent crashes during rapid model switching.

OllamaFX is truly multiplataform (Windows, macOS, Linux).

Download OllamaFX 0.5.0 & Contribute

OllamaFX is a community-driven project. I invite you to download this release and try the local RAG and Vision features.

👉 Download the v0.5.0 Installers here: https://github.com/fredericksalazar/OllamaFX/releases

👉 Visit the Repository: https://github.com/fredericksalazar/OllamaFX

If you find this project valuable, a Star ⭐ on GitHub helps more than you know. We are already planning the 0.6.0 release—what features do you want to see next? Leave a comment below!

OllamaFX the Native & Hardware-Smart Client for Local LLMs

FrederickSalazar — Sat, 27 Dec 2025 17:55:51 +0000

OllamaFX is a native desktop client developed with JavaFX. Its purpose is to provide an intuitive and advanced interaction layer for Ollama, allowing model management and conversations to take place in an optimized and elegant environment.

The project was born from the need for a tool that acts not only as a chat interface but as a complete control center for the models residing on our machines.

Key Features and Benefits

🧠 Integrated Hardware Intelligence

One of the most outstanding innovations of OllamaFX is its ability to understand your system. The application analyzes your hardware specifications and classifies the models in the library according to their viability:

Visual Indicators: You will know in advance which models are ideal for your current configuration thanks to a color-coded system based on resource availability.

Operational Safety: Minimize friction when choosing the right model for each task without compromising system stability.

⚡ Native and Efficient Architecture

OllamaFX leverages the power of JavaFX and AtlantaFX to offer a modern, clean, and extremely fast user interface (UI).

Low Consumption: The application is optimized to be lightweight, ensuring that most of your PC's resources are dedicated to model processing.

Professional Interface: OllamaFX it offers a distraction-free environment with full support for light and dark themes.

🗂️ Session-Based Workflow

Version 0.4.0 introduces a new sidebar designed for multitasking.

Context Management: You can keep multiple sessions open with different models simultaneously.

Persistence: Navigate between your chats with a single click, keeping the history and context of each conversation organized.

📚 Advanced Model Explorer

Discovering and downloading models has never been easier. OllamaFX features a revamped "Home" where you can explore trends, see the most popular models in the community, and manage your local library with a smart caching system for instant loading.

An Open Source Project for the Community 🤝

OllamaFX is a project under the MIT license, meaning it is free, transparent, and open for everyone to collaborate. My vision is to build a community of developers who want to take local AI to the next level.

How can you participate?

Explore and Use: Download v0.4.0 and experience the fluidity of a native tool.

Boost the Project: A ⭐️ on the GitHub repository helps us reach more people and validates the development effort.

Collaborate: The code is open for improvements, new features, translations, and bug reports. Your contribution is vital to the growth of OllamaFX!

👉 Visit the official repository: github.com/fredericksalazar/OllamaFX

OllamaFX is more than just a client; it is a tool designed to empower local LLM users. If you are looking for a native, intelligent, and professional experience, I invite you to join our community.

What functionality would you like to see in future versions? Let me know your ideas in the comments!

DinaWall 0.2

FrederickSalazar — Wed, 15 Mar 2023 00:34:41 +0000

Hello everyone, 3 years ago I started to create DinaWall an app written in JavaFX that implements dynamic wallpapers on Linux, Windows and macOS. After so much time I have decided to continue its development and finish it, which is why I have been working on it for a few weeks. Today it has release version 0.2 and has the following features:

Allows you to install new dynamic wallpapers
allows you to create new dynamic wallpapers
There is a repository on github where dynamic wallpapers are hosted that can be downloaded and installed.
supports the following environments Linux (Gnome, Kde) and macOS, windows is under development

I invite you to get to know the project and contribute to its development, there is a plan of characteristics and functionalities that have the objective of turning DinaWall into a top application in all operating systems, the support of the community and their contributions are essential to be able to continue its development. I invite you to try the application, leave your comments, ideas, report bugs, create your dynamic wallpapers and share them with the community, help with documentation, code, etc. Any contribution is welcome and can help those who start in the world of software to share knowledge and learn from others. I leave the link to the github repo and my email in case you want to write.

Github Repo ->

fsalazars@uoc.edu

Invitation to contribute to the DinaWall project, dynamic wallpapers for the community

FrederickSalazar — Sat, 03 Oct 2020 21:39:45 +0000

Hello, it is a pleasure for me to start my contributions in this community, this is my first post I hope you like it, I take this post to invite you to participate in a project that started recently, when I was using GNU / Linux I wanted to develop an application with which I could implement the macOS Dynamic Wallpapers in distributions Linux, this became a challenge for me, I decided to create the application that I called DinaWall, almost 3 years that I did not write a line of code, since now I am dedicated to business intelligence, but this revived me again.

Let's start with the architecture, I thought to create the application in a totally decoupled way, so I must have 2 components that had to evolve independently, a core which is called DinaWall_core is in charge to manage all system operations, such as file management, wallpapers installation, manage the execution of dynamic wallpapers and take care of the administrative issues of the system, the other component is dinawall_app this is the user interface written in JavaFX, I wanted as a personal challenge to create an application visually attractive and easy to maintain. after two weeks of hard work I managed to meet the objectives raised.

So then we already have a fully functional system, with tests in which the system managed to make the dynamic change of wallpapers according to system time, at this moment, the application supports Linux systems with desktop environment Plasma KDE and I managed to do the implementation for its operation in macOS, there is a lot of work still to be done, implement more Linux desktop environments like gnome, cinnamon etc.

At this time, due to work issues, I have not been able to continue working on the project as I would like and that is why I extend the invitation for you, the community, to help me take this application to a functional level contribute your grain of sand to the open source community, I have many ideas yet to develop for DinaWall and I hope to tell With your support, thank you very much for your attention, it should be noted that documentation is also missing so if someone is interested in contributing to the project you can write to me at this email: fredefass01@gmail.com to coordinate and continue with the project, Thank you again and hugs from Colombia, happy hacking aaaa and excuse my english is very bad

DinaWall_App Github Repository

DinaWall_Core GitHub Repository