Changelog
This is a major update focusing on flexibility, customization, and making VOCR accessible to more users around the world.
There are many internal changes, so we need your help with comprehensive testing to ensure nothing is broken and the new features are working as expected.
Preset System
The old engine selection (GPT / Ollama / LlamaCpp) has been replaced with a flexible Preset system. Each preset stores its own API URL, API key, model, and prompts, so you can set up as many AI configurations as you like and switch between them instantly from the new Presets menu.
API keys are encrypted and stored securely using the macOS Keychain, and are only displayed once in plain text when creating a preset.
A Preset Manager window lets you create, edit, duplicate, and delete presets.
Built-in provider URLs are included for Claude, Gemini, Ollama, OpenAI, and OpenRouter. Any service that offers an OpenAI-compatible API will work.
Follow-up Conversations
A new Follow up checkbox in the Ask dialog lets you carry on a multi-turn conversation with the AI instead of starting fresh each time.
Customizable Explore Prompts
Explore mode uses a structured json-schema response format for more reliable results, and you can now edit the system prompt and user prompt used by Explore mode via Presets > Edit Explore Prompts.
Reset Option
A Reset item in the Settings menu lets you erase all settings and presets and start fresh.
11 Languages
VOCR is now localized in English, German, Spanish, French, Japanese, Korean, Polish, Portuguese, Russian, Ukrainian, and Simplified Chinese.
The translations are done by AI, so we need help from the community to improve. See the translation guide for more information.
Changed
- "Use Last Prompt" has been replaced by "Use Preset Prompt", which sends the prompt saved in the active preset without showing the dialog.
- The Ask dialog button is now labeled "Ask" instead of "Ok".
- AI status messages now show the name of the preset being used (e.g., "Asking Claude... Please wait...").
- Error messages are now shown as alert dialogs.
- Token usage information is now copied to the clipboard along with the AI response.
Removed
- The Engine submenu (GPT / Ollama / LlamaCpp) -- replaced by presets.
- The OpenAI API Key dialog in Settings -- API keys are now managed per-preset.
- The Set System Prompt dialog in Settings -- prompts are now managed per-preset.
Special thanks to @vic08!