Bug Fix 🐛 Fixed an issue where the chat history was not being updated.
Improvements 🔧 Improved the chat search UI.
Library Search 📚 Added tools to search your entire document library in chat.
Alibaba Qwen-QWQ-32B model 🧠 Added Alibaba Qwen-QWQ-32B reasoning model to the supported model list.
Automatic Link Crawling 🔗 Zemith now intelligently extracts information from websites and YouTube videos when you share links in chat. Simply paste a URL, and watch as Zemith automatically crawls and analyzes the content, providing you with relevant insights.
Added file upload in chat as knowledge source You can now upload files in chat and use them as knowledge source.
Sidebar Improvements
Hide Sidebar Email You can now hide the email input on the sidebar.
Hide Install App You can now hide the install app button on the sidebar.
Hide Theme Toggle You can now hide the theme toggle on the sidebar.
Hide Discord Link You can now hide the discord link on the sidebar.
Hide Support Link You can now hide the support link on the sidebar.
Pinned Apps You can now pin apps to the sidebar.
Default Open App You can now select the default app to open.
Perplexity Sonar Model Introducing support for Perplexity Sonar model, which has built-in web search citation.
Perplexity DeepSeek R1:1776 Model
Deprecating standalone search feature We are deprecating the standalone search feature. You can now use a more powerful web search feature within the chat.
Claude 3.7 Sonnet with Reasoning 🧠 Introducing support for Claude 3.7 Sonnet, featuring a toggle for step-by-step reasoning capabilities.
Claude 3.5 Haiku Image Support 🖼️ Claude 3.5 Haiku now supports image analysis and processing capabilities.
Improving Web Search Capabilities 🔍 Added support for automatically generating multiple search queries, enabling more precise and comprehensive web search results.
Expanded Upload Capacity 🚀 File upload limit increased to 50MB, enabling support for larger documents (experimental).
Broader File Format Support 📄 Now supporting an expanded range of formats including PDF, TXT, MD, DOCX, CSV, PPTX, and web pages - improved text extraction capabilities.
Smart Content Analysis 🔬 New automatic analysis feature provides instant insights for all uploaded documents, helping you understand your content faster.
Enhanced Deep Dive Conversations 💬 Improved conversation engine delivers more comprehensive and engaging conversations.
Advanced RAG Search 🔍 Significantly improved search accuracy with new preprocessing pipeline, making your document queries more precise and relevant.
Expanded Model Selection 🧠 Added additional AI models to the library chat.
Fix CodeBlock Render in Chat ⚙️ Resolved code block rendering issues with o3-mini output for better readability.
Fix text rendering issue 🤖 Fixed the issue where some text formula was not rendering properly.
Fix Google Imagen 3.0 text-to-image generation 🧠 Fixed the issue where Google Imagen 3.0 text-to-image generation was not working.
Display used model in assistant message 🤖 You can now see the model used in the assistant message.
Qwen-2.5-vl-7b model 🧠 Added Qwen-2.5-vl-7b model to the supported model list, this model usage is unlimited for all upgraded users.
Hide email toggle on sidepanel
Default Model Selection 💬 You can now select the default model for your chat.
Introducing Mistral Model Added Mistral large models to the supported model list.
Display Credit Cost in Model Selector
Message Regeneration Feature 💬 Not satisfied with an AI response? You can now regenerate any assistant message to get alternative answers with different models. This gives you more control over your conversations and helps you get the perfect response every time.
Fixed iOS sign in issue Fixed an issue where iOS users with Safari browser were unable to sign in occasionally due to popup blocker
Added dedicated page for Creative AI Results Share your Creative AI generations easily with a dedicated results page that anyone can view.
DeepSeek-R1 Model Enhancement We've upgraded our DeepSeek-R1 model to a US-based infrastructure, delivering faster response times and enhanced data privacy for our users.
Microsoft Account Linking We've added Microsoft account linking to the platform.
Fix video generation not retrieving results Fix the issue where video generation was not retrieving the generated video.
Updated the default settings for video generation
Document to Podcast tool Introduce Experimental Custom Document to Podcast tool.
Ability to export quizzes You can now export quizzes to a HTML file so that you can use them offline.
Auto-Save for Generated Quizzes Quizzes created with the Document-to-Quiz tool are now automatically saved.
Update monthly credit refresh to reflect the name of the month
Discord link added Thanks to community member Alexander for setting up the Discord server!
Fix text formatting issue in notepad
Default System Prompt Feature Set a default system prompt that automatically applies to all your new chats, streamlining your interactions.
Zemith is now available as a PWA! 📱 Install Zemith directly on your Android, iOS, or desktop device for a seamless app-like experience. Get quick access to Zemith right from your home screen while we continue developing our native mobile apps.
Enhanced Sidebar Navigation 🔄 We've made the sidebar scrollable, ensuring you can access your full chat history even on smaller screens. This improvement makes navigation more fluid and prevents important conversations from being cut off.
OpenAI o3-mini-high model We have added OpenAI o3-mini-high model to the supported model list.
Document to Markdown tool We have added Document to Markdown tool to the supported experimental tool list.
Google Gemini 2.0 Flash Lite & Gemini 2.0 Flash We have added Gemini 2.0 Flash Lite & Gemini 2.0 Flash to the supported model list.
Default model changed to Gemini 2.0 Flash We have changed the default model to Gemini 2.0 Flash.
Fix Text Formatting Issue We have fixed the issue where text formatting was not rendering properly.
Account linking You can now connect your Zemith account to other social platforms. This makes it easier to sign in with your existing accounts and helps us prepare for our mobile app launch.
Date awareness for LLM The LLM now knows the current date. This makes it useful to retrieve current events or news during web search.
Text-to-Speech for Chat Assistant Responses You can now listen to chat assistant responses using text-to-speech functionality. Currently available in English only.
Dark Mode Update Updated the dark mode color scheme
OpenAI o3-mini and o1 models You can now use OpenAI o3-mini and o1 models in chat.
Alibaba Qwen-2.5-max models You can now use Alibaba Qwen-2.5-max models in chat.
Web search improvement Improved the web search to be more accurate and relevant.
Message length limit Increased the chat message limit. For messages beyond 50,000 characters, the credit usage will be doubled.
Introducing Custom System Prompt You can now customize the system prompt for your chat.
Introducing Prompt Gallery You can now share your prompts with the community.
Introducing chat search You can now search across your chat history to find specific messages.
Introducing Dark Mode Dark Mode has arrived. Now you get to choose between light and dark mode.
Improving accessibility We also added some minor tooltips for better accessibility.
Introducing DeepSeek v3 and DeepSeek r1 models DeepSeek r1 shows its step-by-step reasoning process, making its thought process transparent to users.
Introducing roadmaps and changelogs to Zemith.
We are introducing a roadmap tracker and changelog to the platform to share the progress we are working on with the community.
Enabling web search for all models Zemith paying users now have access perform web search with all models in chat.
You can now enable web search for all models in chat settings.