FlowDown is a simple and fast AI chat app designed to make working with large language models easy and effective. It helps you have focused and smooth conversations while giving you control over your interactions.

One of the best things about FlowDown is that it is lightweight and efficient. This means it launches quickly and lets you focus on your conversation instead of getting distracted by the app itself. Even during long chats, FlowDown remains responsive, making it perfect for users with busy machines.

FlowDown also supports Markdown formatting, ensuring you get clear and readable answers. It beautifully renders Markdown, so you can easily scan code blocks, bullet lists, tables, and headings in the AI responses. When it comes to connecting with the tools you already use, FlowDown works well with any OpenAI-compatible API. This flexibility allows you to connect to major cloud providers or even services running in your own home lab.

You can interact with FlowDown using text, images, and audio. This feature is beneficial for tasks like summarizing voice notes, transcribing meetings, or asking questions about images. The app allows you to show, rather than just tell, making your interactions richer and more engaging.

FlowDown provides a smooth, instant-feeling chat experience. The app’s fast text rendering means responses come through fluidly, so you can read and refine your messages without needing to wait for the model to finish generating replies.

To help you manage your conversations easily, FlowDown features automatic chat titles. The app names your chats based on their contents, making it simple to find discussions later, whether they are about an "API bug fix" or "trip planning."

With Shortcuts support, FlowDown allows you to integrate the app into your daily workflows. You can send selected text from other apps to your model, generate text drafts or replies with a single tap, and even build custom automations that call your favorite models. This makes the app highly versatile for your needs.

Privacy by design is a key focus of FlowDown. Your data remains yours, and the app does not collect your chats. For those seeking maximum confidentiality, you can run models locally on your Apple Silicon devices without needing to connect to external servers.

FlowDown is open source, and you can access the entire codebase on GitHub. This transparency allows you to inspect, audit, or customize the app for your own needs. We also offer a free trial, giving you the opportunity to see if FlowDown fits your workflow before making a purchase. You can find the source code and trial versions through a simple visit to our GitHub page.

For offline AI functionality, FlowDown can run models locally, allowing for a secure and private experience. Supported models include the Cohere model, Gemma model, OpenELM model, Phi model, and Qwen2 model. This feature gives you full control over your data without any dependency on external servers while allowing you to experiment with various open models.

FlowDown also includes remote models, providing a complimentary service to start chatting right away. However, this starter service does come with rate limits and no uptime guarantee. For reliable use, we suggest bringing your own API or hosting your own inference server on local or home lab hardware. The app supports all OpenAI-compatible endpoints, making it easy to connect with tools like LM Studio, enhancing your privacy and options.

Keep in mind that some setups require a bit of technical knowledge. If you are new to self-hosted AI or custom endpoints, we recommend reading the documentation carefully, as it can be challenging without it. We strongly suggest trying the free trial to ensure FlowDown meets your needs before purchasing since refunds may not be available.

Lastly, please remember that AI-generated content can sometimes be inaccurate or misleading. It’s important to verify crucial information and use FlowDown responsibly, as you are responsible for how you utilize its outputs.