Changelog
Follow up on the latest improvements andΒ updates.
RSS
π New
- Added GPT-4o 20241120 for OpenAI.
- Added Sonar Huge Online (405B) for Perplexity services.
- Added Ministral (3B, 8B), Pixtral (Large, 12B) for MistralAI services.
- Added Gemini Flash 1.5 8B (Latest, Stable), Gemini Pro 1.5 Stable (Gemini Pro 1.5 002), Gemini Flash 1.5 Stable (Gemini Flash 1.5 002) for Google Gemini services.
- Added LLaMA 3.1 70B Specdec, LLaMA 3.2 Preview (1B, 3B), LLaMA 3.2 Vision Preview (11B, 90B) for Groq services.
- Added Gemini Pro 1.5 002, Gemini Flash 1.5 002, Claude 3.5 Sonnet v2, Claude 3.5 Haiku for Google Vertex AI service.
- Added Claude 3.5 Sonnet v2, Claude 3.5 Haiku for Anthropic Claude services.
- Added xAI support with Grok and Grok Vision models.
π§ Fixed
- Resolved the problem when adding new Anthropic Claude and Google Vertex AI endpoints.
π New
- Introduced new OpenAI models: o1-preview and o1-mini. Please notice that these models are exclusive to Tier 5 users who have spent over $1,000 and 30+ days since first successful payment.
- Added new models, including LLaMA Guard 3 and LLaVA 1.5 to Groq service. Users can now upload images and ask questions with the LLaVA 1.5 model on Groq.
- Added support for Cerebras models, further expanding our AI capabilities.
π§ Fixed
- Resolved an issue preventing the addition of new Google AI endpoints.
π New
- Added GPT-4o Long model for OpenAI.
- Added LLaMA 3.1 support for Groq and OctoAI services.
- Added LLaMA 3.1 (Llama 3.1 API Service) support for Google Vertex AI service.
- Updated models and removed deprecated models for Perplexity service.
- Added the ability to archive and unarchive multiple conversations and folders.
- Introduced a new keyboard shortcut, Command + Shift + C, to copy the last response of a conversation.
- A trash bin has been added, where deleted conversations and folders will be stored before being permanently deleted after 30 days.
β¨ Improved
- Enhanced response handling for OctoAI.
- Set default max tokens to 4096 for Anthropic Claude models.
π§ Fixed
- Auto spell checking now works correctly in the prompt editor.
- Resolved the problem where the prompt editor's height was not resetting after sending with quick action buttons.
π
New
- Added support for GPT-4o Mini for OpenAI, includes vision capabilities.
- Added support for Mistral NeMo & Mistral Codestral Mamba.
π§
Fixed
- Fixed an issue where only the first character of the prompt was submitted when pressing the send button or quick action buttons.
new
improved
fixed
MindMac 1.9.18: GPT-4o on OpenAI Azure & some improvements
π
New
- Added support for GPT-4o on OpenAI Azure, includes vision capabilities.
β¨
Improved
- Enhanced overall scrolling experience for better user interaction.
π§
Fixed
- Resolved an issue where the prompt editor failed to display prompt templates when selected from the built-in prompt library.
π
New
- Added Gemini 1.5 Flash model and Gemini 1.5 Pro with 2M context length.
- Introduced Anthropic Claude 3.5 Sonnet and Mistral Codestral models.
β¨
Improved
- Removed the delete button from the top-left corner to prevent accidental deletion of all conversations.
π§
Fixed
- Fixed performance issue when typing very long prompts.
- Fixed issue max_tokens is too large when use Anthropic Claude 3 models.
- Fixed issue where the prompt text was not cleared after being sent to the AI provider.
- Resolved issue with adding Vertex AI at the us-east5 region. Now, when adding the Vertex us-east5 endpoint, use claude-3-opus@20240229 as the input model.
- Fixed issue where the system and user roles were not alternating correctly with Perplexity models when the first message was from the assistant and context mode was set to the last 5 messages.
new
improved
fixed
MindMac 1.9.16: GPT-4o, audio features & better file upload
π
New
- Added support for OpenAI's new flagship GPT-4o model with vision and text capabilities.
- Introduced DeepSeek models.
- Added new models while removing deprecated ones for Perplexity.
- Implemented Speech to Text feature with support from both OpenAI and Apple native capability for prompt input.
- Added support for Text to Speech, allowing users to choose between OpenAI and Apple's native capability for reading AI responses out loud.
- Added actions for file upload, allowing audio files to be uploaded for transcription, summarization, translation, and key point extraction.
- Added support for copying and pasting files (image, audio, document) into prompt editor.
β¨
Improved
- Updated app Settings UI.
π§
Fixed
- Resolved occasional scroll issue causing jumping.
- Resolved issue of receiving infinite responses from Ollama llama2 model.
- Addressed problem with stream response not functioning well with Perplexity models.
- Fixed issue where the message view was being overlapped when the prompt editor grew in size.
π
New
- Added support for Llama 3 models on Anyscale.
- Added functionality to enter the model name during the validation of the Vertex AI endpoint.
β¨ Improved
- Enhanced the scrolling effect for smoother navigation.
- Updated the user interface for the message editing view.
π New
- Added support for Llama 3 models on Groq and Perplexity.
- Introduced support for LaTeX formatting.
β¨ Improved
- Optimized overall application performance, including improved scrolling, better handling of long messages, and simplified multi-paragraph selection for copying.
- Disabled network proxy when using local inference sources such as Ollama, LMStudio, MLX.
- Transitioned explanatory messages to popover form in specific settings fields.
- Enhanced user interface for the main message view.
π§ Fixed
- Resolved problems preventing some users from utilizing Claude 3 models on Vertex AI.
π
New
- Added custom model field for Ollama when creating new endpoint.
β¨
Improved
- Updated model name for Cohere Command R+.
- Included service logos next to each answer from AI.
- Enhanced document upload capabilities, specifically for Excel and CSV file formats.
- More detailed error messages now shown for Vertex Claude 3 models.
π§
Fixed
- Resolved a bug causing crashes during document uploads.
- Addressed a crash that occurred during the cleanup of conversations for certain users.
Load More
β