Mindute AI adds Roundtable, a new way to compare how different models respond to the same prompt. Roundtable lets you bring multiple models into one conversation-style view, making it easier to compare reasoning, tone, creativity, and usefulness side by side. This release also adds a clearer model load indicator, so it’s easier to see when a model is preparing, loading, or ready to chat. Also included: - Improvements to model loading feedback - UI polish and reliability improvements - Bug fixes and stability improvements throughout the app

Mindute AI gets a refreshed, smoother experience across chat, models, settings, agents, and the sidebar. This release adds a much better chat history experience: you can now search, scroll, pin, rename, archive, share, and delete conversations directly from the left navigation drawer. Model browsing has been redesigned with clearer source views, improved search and filters, and a smoother first-run model recommendation flow. Advanced users can also opt in to bring-your-own-key cloud models through OpenRouter, with dynamic model discovery and direct provider requests. Privacy remains central: local models stay on your device, OpenRouter API keys are stored on device, and Mindute servers never receive your API keys. Also included: - Experimental: Live voice chat using local models which should work like a treat - Cleaner chat loading and writing states - Better model and agent selection flows - UI polish and reliability improvements throughout the app

Mindute AI gets a refreshed, smoother experience across chat, models, settings, agents, and the sidebar. This release adds a much better chat history experience: you can now search, scroll, pin, rename, archive, share, and delete conversations directly from the left navigation drawer. Model browsing has been redesigned with clearer source views, improved search and filters, and a smoother first-run model recommendation flow. Advanced users can also opt in to bring-your-own-key cloud models through OpenRouter, with dynamic model discovery and direct provider requests. Privacy remains central: local models stay on your device, OpenRouter API keys are stored on device, and Mindute servers never receive your API keys. Also included: - Improved Settings screen with system/light/dark appearance options - Cleaner chat loading and writing states - Better model and agent selection flows - Backup restore from file or clipboard - Safer cloud error messages without exposing URLs, response codes, or key details - UI polish and reliability improvements throughout the app

• Hugging Face Marketplace — browse, search, and download community models without leaving the app. • On-device Image Generation — create images right in chat using Apple Foundation Models, with multiple style options (Animation, Illustration, Sketch) to choose from. • A more refined chat experience with smoother message rendering and small UI improvements throughout.

Mindute AI 3.0.0 is a major upgrade focused on better models, smarter local tools, and a smoother private AI experience. • Browse models more easily by task or brand • Added clearer model availability guidance for your device • Added support for more on-device model families, including Apple Intelligence on supported iPhones • New local tools for calculations, unit conversion, word count, translation, OCR, clipboard actions, and Reminders • Improved vision and document handling • More reliable model downloads with better progress, retry, and validation • Stronger chat and agent storage reliability • Cleaner chat experience with better formatting and tool result display • Performance, stability, and test coverage improvements throughout the app

We’ve made Mindute AI faster, smarter, and more intuitive than ever: Fresh new design: A completely redesigned interface that’s cleaner, smoother, and easier to use. More powerful AI models: Access a wider range of on-device models, including: • Apple Foundation Models (on supported iPhones) • Gemma 4 models (on selected devices) • Llama 3.2 models • Qwen 3.5 models Smart model recommendations: Automatically suggests the best models based on your device’s RAM for optimal performance. Better model organisation: Models are now grouped by tasks (e.g. chat, coding, vision) to help you find the right one faster. Multilingual responses: Choose your preferred language for AI responses — with more languages coming soon. Smarter chat experience: Improved chat organisation and management for a more seamless experience.