- The preset list of models has been updated, adding gpt-4-turbo-preview model and removing gpt-4-1106-preview and gpt-4 models.
- Max Tokens setting in the Options has two possible settings now: Auto-set and Custom. For most cases, it is recommended to use Auto-set.
- Previously, when the Max Tokens setting was set to a value that exceeded the Context Window limits for the selected model, the user saw a confusing message instructing them to shorten the prompt length. This is now fixed.
- All keyboard shortcuts are now displayed in uppercase and use a more consistent format.
- In the Options window, the Frequency and Presence Penalty settings now have a range from -2 to 2 for consistency with OpenAI documentation.
- In the Shortcut List window, the list of prompts associated with keyboard shortcuts is now sorted by the keyboard shortcut by default, and it is possible to sort by other columns in ascending and descending order.
- Slight UI adjustments have been made to the Send buttons in the main window: the text of the top Send button has been changed to “Send New”, the layout has been changed, and the tooltips of both Send buttons has been modified.
- A new menu item called “OpenAI Developer Platform” has been added under the Help menu in the main window.
- It is now possible to undo recent changes in the Instructions and Segment text fields if these fields are cleared using the Clear Instructions And Segment or Hide Dialog History / Clear Instructions And Segment buttons.
- In the Prompt Library, double-clicking a prompt now loads this prompt in the main window, similar to the Load Prompt button.
- Several minor fixes and improvements.
|February 27, 2024
|Fixed a bug in the Options window which affected some users.
|December 28, 2023
- The Prompt Library has been significantly revamped, providing you with a single point for managing all of your AI prompts. The Prompt Library is accessible from the Prompts button in the main window or from Tools > Prompt Library menu item. Its features include: 1) Ability to create, update, rename, copy/move and delete prompts. 2) Ability to create, rename and delete folders in order to organize prompts into a convenient folder structure. 3) Easy configuration of custom prompt settings – keyboard shortcut, shortcut action, AI model and temperature. 4) Remembering of the last expanded/collapsed state of each folder to help you find often-used prompts. 5) Favorites feature that makes it easy to create a list of favorite prompts for quick access. 6) The Prompt Library window can remain open alongside the main CotranslatorAI window so you can manage your prompts while using CotranslatorAI.
- Added the “gpt-4-1106-preview” model (early name of the new GPT-4-Turbo model) to the quick list of models. To use this model (if you don’t already have access to the GPT-4 models), you must add a credit card under your OpenAI API account (https://platform.openai.com/) and you must be invoiced or have paid for a small amount of API usage. For example, you can do so by purchasing credits under your OpenAI API account.
- Renamed “CotranslatorAI” to “CotranslatorAI Free” as the official name for version 2.0 of the application.
- Added links to online resources under the Help menu.
- Small bug fixes in the Options dialog.
- Minor UI and usability improvements across the application.
|November 27, 2023
- Copy to Clipboard button was added in the Prompt Library window, making it easier to use CotranslatorAI’s Prompt Library with various AI services.
- Minor UI and usability improvements.
- CotranslatorAI no longer prevents the user from invoking built-in keyboard shortcuts from inside CotranslatorAI window (previously the following dialog was displayed: “This keyboard shortcut cannot be activated from the CotranslatorAI dialog. You should run CotranslatorAI prompts from other applications.”). This makes it easier to run advanced prompts such as variable-driven prompts, multi-input prompts, scripted prompts, etc.
- Fixed a few bugs.
|October 15, 2023
- The dialog history which displays the “conversations” with the AI has been revamped:
1) Dialogs that share the same context are now separated with headings, making it easier to navigate;
2) Tab characters are now displayed using long arrows similar to the way they are displayed in Microsoft Word when non-printing characters are made visible;
3) It is more convenient to navigate the dialog history when there are very long dialogs.
- It is now possible to change CotranslatorAI settings, including OpenAI settings, using the new CotranslatorAI Options window accessible via Tools > Options menu. Therefore, it is no longer necessary to edit the configuration file directly in order to change the API key or any other settings.
- The size of Prompt Library window and the position of the vertical divider in this window are now remembered.
- More readable error messages are displayed in the event of OpenAI API errors.
|September 20, 2023
- CotranslatorAI’s main window was not being activated by keyboard shortcuts if it was minimized. This has now been fixed.
- “SendOnly” shortcut action setting has been renamed to “TransferOnly” both for prompt files and for he configuration file. “SendOnly” is still supported and is equivalent in meaning to “TransferOnly”.
- Other minor fixes.
|September 4, 2023
- Introduced a new prompt file setting called Shortcut Action. This setting controls what happens when you invoke the keyboard shortcut associated with a prompt file. The default setting is Paste Response, i.e. copy the prompt text to the Instructions field, copy the selected text to the Segment field, submit a query to the AI, receive response and paste the response over selected text. There are two other possible values for this setting: Send Only fills the Instructions and Segment fields without submitting an AI query, and Get Response fills these fields and submits an AI query without pasting the AI response over the selected text. Syntax: [ShortcutAction = PasteResponse / GetResponse / SendOnly]
- It is now possible to specify the default Shortcut Action for prompt files by adding the appropriate setting in the application’s configuration file. Syntax: prompt-shortcut-action = PasteResponse / GetResponse / SendOnly.
- Added a token counter in the application’s status bar. When you paste or type text in Instructions, Segment or Ad-hoc Prompt fields, the application will tell you how many tokens will be used up when a prompt is submitted to the AI engine. This is useful for making sure that you do not exceed the limits of the selected AI model and the AI response is not truncated.
- Added a Token Counter tool under Tools menu. This tool can be used to determine the number of tokens in a specific prompt, to make sure that model’s token limits are not exceeded, e.g. when creating prompts for your prompt library.
- After an AI response is received from the AI engine, the application will indicate how many tokens were submitted to the AI, how many were received in the response, and the total token count in the request and response added together. These details will have the following format: Previous — Request Tokens / Response Tokens / Total Tokens.
- Removed GPT-3 (text-davinci-003) model choice from the dropdown list in the main window.
- The Prompt Library dialog can now display a subfolder of the Prompts folder if it is empty or contains no valid files.
- Updated terminology across the application for clarity and consistency.
- Removed Final Marker setting in the Prompt Library window as this setting is pertinent to legacy models such as text-davinci-003.
- It is now possible to open the configuration file from the application by clicking Tools > Configuration File menu item.
- Minor fixes.
|Aug 21, 2023
- The user interface of CotranslatorAI has been significantly revamped. All command buttons have been moved to a single compact area (referred to as the “Dashboard”) in the top right corner of the main dialog, making it easier to click the right button.
- There are now two different commands for clearing the chat history. The default command accessible from the Dashboard is called Hide Chat History. This command hides all previous chat history, but does not delete it, so that the hidden chat history can be used when sending ad-hoc prompts from the Ad-hoc Prompt area at the bottom of the chat panel. The second command called Flush Chat History deletes all previous chat history, including hidden chat history, so that there is no previous context for any prompts. Flush Chat History command is available from the Tools menu.
- To send an ad-hoc prompt from the Ad-hoc Prompt area at the bottom of the chat panel, you should now click the new send button next to the text entry box (instead of the button in the button area as was the case in previous versions), or press Ctrl+Enter.
- Several minor usability improvements in user messages and UI.
- When Increase Font Size menu item is used, the dialog’s size is retained, similar to the way the Decrease Font Size menu works.
- When Copy Chat History button is used, the user now gets a choice to copy either the full chat history (including hidden chat history) or only the chat history visible in the chat panel.
- It is now possible to cancel prompts before the AI response is received, using the Cancel button.
|July 26, 2023
- If shortcut keys are used to process selected text which contains trailing breaks, these breaks will now be preserved when the AI response replaces the selected text. There are some exceptions to this for Word users, see User Best Practices page at https://cotranslatorai.com/user-best-practices/ for more details.
- Fixed several bugs which prevented CotranslatorAI from working correctly when predefined shortcut keys or prompt shortcut keys were used.
- Fixed a UI bug which occurred if Decrease Font Size menu item was used while the CotranslatorAI window was maximized.
- Ctrl+Shift+WinKey+F12 was not recognized as a predefined shortcut key in the Shortcut List dialog or when checking shortcut keys of prompt files being loaded.
- When Decrease Font Size menu item is used, the window size is now retained, allowing user to see more text. Previously, the window size was reduced.
- Improved the stability and user experience.
|June 21, 2023
- AI response was not always pasted back over the selected text when predefined shortcut keys or prompt shortcut keys were invoked. This is now fixed.
- Added a new predefined shortcut key: CTRL+SHIFT+WIN+F12. This shortcut key copies the text selected in any application to the Segment field of CotranslatorAI and sends the entire prompt to the AI, but does not overwrite the selected text with the AI response, unlike CTRL+SHIFT+ALT+F12 shortcut key.
- The ad-hoc prompt label at the bottom of the chat panel (where an ad-hoc prompt can be entered) was changed from “Prompt” to “Ad-hoc Prompt” for clarity.
- If CotranslatorAI is not yet running and a predefined shortcut key or a prompt shortcut key is invoked, CotranslatorAI will be started in order to execute the action. Consequently, there is no need for you to start CotranslatorAI application before running prompts using shortcut keys associated with prompts. When predefined shortcut keys are invoked and CotranslatorAI is not yet running, once CotranslatorAI window is displayed, you will see a notification about missing instructions.
- Fixed several minor bugs.
|June 14, 2023
- A readable message is now displayed if the OpenAI server is taking too long to respond to a prompt. Previously, a timeout error was displayed in such cases.
- Various fixes to improve the stability and user experience.
- Pause Script menu item was removed from the right-click tray menu of CotranslatorAI Companion. Use Suspend Hotkeys menu item if you want, temporarily, to prevent CotranslatorAI from processing all shortcut keys in order to use some of these shortcut keys in another program.
- The ad-hoc instructions row at the bottom of the chat panel now uses a white background color to differentiate it from other rows.
- While the application is waiting for the response from the AI, you will now see progress notifications in the Windows task bar at the bottom of the screen.
|May 29, 2023
- This first public release requires a new free registration code. To receive it, after installing the new version, click “Register for free” button when you start CotranslatorAI and follow the instructions to get your free registration code.
- It is no longer necessary to select text in order to use Ctrl+Shift+Alt+F12 shortcut key (which sends the current prompt + selected text to the AI, receives the response, and places the response over the selected text) or shortcut keys associated with your own prompts. This allows creating prompts that insert AI-generated or fixed text instead of processing selected text. So, if a prompt does not require “Segment” text, simply place the cursor anywhere and activate the shortcut key in order to insert text at this position.
- Added a new menu item: Tools > Shortcut list. Use this command to view the list of shortcut keys, including those associated with your prompt files, and to check whether there are any issues with these shortcut keys, e.g. invalid key combination, duplicate shortcut key, usage of reserved shortcut keys, etc.
- Upon loading prompt files, CotranslatorAI Companion now provides an error report about issues with shortcut keys associated with your prompts (e.g. invalid key combination, duplicate shortcut key, usage of reserved shortcut keys, etc.).
- The application can now handle larger prompts.
- Improved readability of error messages: various error messages returned by the AI are now easier to understand and act upon.
- Upon installation, an empty “My Prompts” subfolder is created in the Prompts folder. It is recommended to place all custom prompts within this subfolder.
- Only one instance of CotranslatorAI application can now be run simultaneously.
- Dividers in the main CotranslatorAI dialog and in Select Prompt dialog are now visible, making it easier to adjust the size of panels in these dialogs.
- Fixed a bug with recognition of numeric keys 0-9 as part of prompt shortcut keys.
- Fixed several miscellaneous bugs.
|May 4, 2023
|April 11, 2023