Thinking support - specialized AI models can now spend more time thinking before they respond. It uses Fluid Pro's DeepSeek R1 distilled Llama 3.3 70B but it is capable of working with local models too.

Fixed chat performance bug in thread view.

Ability to disable Sentry reporting in Settings.

Copy changes
 

Ability to connect to external providers (ChatGPT, Claude, Ollama and others)

Ability to purchase Fluid Pro or license straight from the Fluid app

Ability to turn off auto-update and/or auto-install

Improved FluidPro error handling

Dedicated fluid account tab in settings

Redesigned app's menu - it's smaller now and more readable

Fluid Pro is available for everyone! No matter what Mac you have, you can now use our super capable 72B AI model (Qwen 2.5 or Llama 3.3). We host it ourselves, on our own hardware, making sure it's 100% private 🔐 and powerful 🚀.
Read more about our extreme zero-logging/zero-storage policy.

Fix issue with app expiration.

Ability to select Fluid's writing style: Logical (e.g. for puzzles or math), Balanced (default) and Creative (for writing or brainstorming)

⌘N shortcut now switches to a new thread

In "Light" appearance mode text selection color is visible now

When thread is too long a sign 🐢 is displayed warning about potential quality drop

Bug with Fluid repeating its output infinitely should be mitigated

Fixed bug where Fluid was unable to load very long conversations

Other bug fixes and UI improvements

Llama 3.1 brain with support for Spanish, Portuguese, Italian, German, Thai, French and Hindi languages (in text chat)

Full chat history

AI search for chat history (e.g. if you type "Sahara", messages mentioning "desert" will be returned).

Ability to start new chat / reset conversation

Ability to load and continue any of the past conversations

Ability to "pin" window so it always stay on top

Rebuilt message rendering engine with code highlighting and ability to select arbitrary chunks of text.

Other bug fixes and UI improvements