Geeps Version 1.4

New features

• New provider management - add multiple OpenAI compatible providers that are synced across devices with iCloud
  • Most provider settings are synced with iCloud (except API keys, iCloud keychain support will be added later)
  • You can add multiple instances of the same provider and assign a different default model and prompt to each instance
  • Enable / disable provider without deleting it
  • Change model settings per provider (like temperature, max tokens, etc) - this will be updated soon to support additional settings like reasoning effort/output and others
• Prompt library v1 - save multiple prompts, synced with iCloud
  • Assign default prompt per provider in API settings
  • Replace prompt for chat (via Chat Settings)
• Load models directly from the API - get new models as soon as they are available in the API
  • the API must support /v1/models endpoint, hard-coded models will no longer be used.
  • If the required model is not in the list you can still enter it manually
  • Models need to be loaded on each device (model list not synced)

Improvements and bug fixes

• Added built-in Google Gemini API support (via OpenAI compatibility mode)
• Improved error messages, returning http codes for clarity.
• Filter OpenRouter models to only show free ones
• Groq endpoints are now supported (including model loading)
• Use local models exposed via LM Studio on localhost or over the internet with Tailscale
• Added gpt-image-1-mini model
• Fixed a crash when downloading generated image on iOS 26 / macOS 26
• When using 'Select text' option on chat message the sheet no longer dismisses when opened for the first time
• Fixed an issue where copy code button was not always working in code blocks
• macOS: when conversation is deleted, the next one is selected automatically
• macOS: updated macOS Settings window for easier navigation and layout
• Performance improvements, reduced project dependencies

Deprecation notice

• The built-in OpenAI Compatible Provider will be removed in the next update! Please use the new 'Add Custom Provider' option instead.
• on visionOS the native version will be replaced by an iPad version

Wait, what about Liquid Glass!?

Liquid Glass and iOS/macOS 26 support is coming soon! In fact this version 1.4 is already built with the new iOS/macOS 26 SDK and the upcoming update will bring refreshed design while still supporting iOS 18 and macOS 15. I'm aiming to ship it with v1.5. Stay tuned!

Previews

Swipe through the images below to see feature screenshots.

1/3Add multiple custom providers, synced with iCloud

2/3Prompt library v1

3/3Improved provider settings, load models from API, enable / disable providers, assign default prompt