Version 0.0.18 is here, bringing several new features and bug fixes 
Highlights
- New: Added support for over 20 new providers! Learn more about the AI Model Directory and how it’s powering the AI models available in AgentOne.
- New: Chats longer than 100 messages are now virtualized by default to improve performance. In our tests, this made some chats load up to 38x faster.
- New: Built-in support for Ollama and LMStudio has arrived! When you first install AgentOne, it will automatically detect and configure these providers for you.
- There are now over 8000 built-in extensions!
- Reworked provider settings with separate Built-in Providers and Custom Providers sections.
- Added model metadata editing for built-in and custom providers, including custom context window, token limit, and pricing overrides.
- Added a “Find dangling extensions” tool to the extensions filter menu for cleaning up leftover installs.
Miscellaneous Changes
- Lots of small but meaningful tweaks!
- OpenAI-compatible providers that did not need their own SDK package are now defined inline in the provider registry.
- Simplified the extensions filter menu: replaced the All/Installed/Built-in selector with a single “Only show installed” option.
- Custom and installed extensions are now sorted to the top of the extensions list by default.
- The chat usage indicator now shows
$?(or≥ $X.XXfor partially-priced chats) instead of$0.00when pricing data is missing for one or more models used in the chat. - Shift-clicking a chat now selects it.
- Various UI and performance improvements.
- Fixed a bug where extensions could appear uninstalled after a registry update while still being installed in the background.