This maintenance release lets you mirror DEVONthink reminders to Apple Reminders via a new script. It also includes a hidden preference to disable dark mode only for PDF documents on macOS Tahoe or later, and lets you run batch processing on specified records via AppleScript. The Pro and Server editions support Claude Sonnet and Opus 4.6 and improve the support for zAI and Qwen models. The update brings several fixes, too, e.g., for lists in rich text documents not displaying in macOS Tahoe.

DEVONthink introduces a refreshed look with a new app icon and full support for Apple's Liquid Glass design on macOS Tahoe. On macOS Sequoia and later, sidebars and inspectors now animate smoothly in and out, and the Navigate sidebar is spring-loaded. Simply drag items to the window edge to reveal it automatically. PDF documents on Tahoe also respect your dark mode preference for more comfortable reading.

The Chat assistant gains powerful new capabilities: adjust many options directly within the chat window, and access up to 50 saved conversations to pick up right where you left off. You can now choose between DEVONthink, Perplexity, or Exa for web and Wikipedia searches, giving you control over speed versus freshness. Updated AI models including Claude 4.5, GPT 5.2, Gemini 3, and Google Nano Banana Pro, deliver improved performance across all AI use. When generating images, you can choose the aspect ratio.

New workflow tools help you work faster: the detachable Graph popover lets you visualize document connections, you can manually save named versions for finer revision control, and annotation files are now easier to create, locate, and open directly from menus and inspectors. Markdown documents also support smart quotes and dashes for polished typography.

This release helps mitigate AI risks, especially when processing online content, to reduce the potential for malicious prompt behavior. We also updated the supported AI providers, adding support for Google's Nano Banana image engine and Anthropic's Claude 4.5 Sonnet.

DEVONthink now detects empty pages when clipping PDFs, ensuring more consistent and useful results, and we made the Safari and share extensions more reliable on macOS Tahoe. Mailboxes in the Import sidebar load faster and match Apple Mail's appearance and you can modify a document's searchable text via the chat or scripting. On the sync side, DEVONthink now automatically adjusts resource usage based on your machine's temperature and workload, and sync in general is more performant. Of course, we fixed bugs and glitches, too.

Here it is, the first feature release for DEVONthink 4. Choose from a much wider range of server-based AI models via OpenRouter or by connecting to just any OpenAI-compatible provider. Integrate models from, e.g., Amazon, Meta, Microsoft, xAI, Baidu, Moonshot AI, Qwen, and more. In addition, DEVONthink Europa adds native support for Claude 4.1 Opus, Gemini 2.5 Flash Lite, OpenAI GPT-5, Google Imagen 4, and the latest models by Mistral. Connect to your institution's Ollama installation with a new configuration option.

To further automate your work, describe your task to the the AI-powered script assistant. It generates AppleScript and JavaScript for Automation (JXA) scripts and even corrects common scripting issues. Edit images, both generated or dropped in from other sources, with prompts using common AI image generation models. As generative AI is notoriously bad at maths, the Chat and other AI-assisted functions now support mathematical functions, and if you're working in academia, retrieval-augmented generation (RAG) gets direct access to the huge document collection on arXiv. In addition, DEVONthink Europa comes with a new Quick Look extension for previewing Spotlight results on macOS Sequoia and later, and more responsive editing of especially larger Markdown documents. It also updates and sorts lists and empties the trash significantly faster.

This maintenance release introduces support for Google Imagen 4 (Preview), along with several improvements and fixes. We enhanced handling of relative links in Markdown documents to allow spaces, and we corrected an issue with list indentation. We also improved document language detection, which can influence Chat-suggested properties like tags. The Favorites section, e.g., in the Move To context menu now correctly lists all favorited groups. Colored groups also show their designated colors in more locations, such as the Path bar. For AI features, we expanded the prompt field in the Chat and Help assistants to accommodate longer prompts. We extended the timeout period for image generation, and queries will retry up to two times when necessary. As always, we focused on increasing the application's stability and reliability.

Our fourth major release — DEVONthink 4.0 Copernicus — brings a host of powerful new features and improvements. DEVONthink now supports external AI providers, both commercial and local AI, as a complement to its own internal engine and implemented with privacy and data safety in mind. You can also write search queries in natural language, let the AI summarize or transform text, and extract text from images or handwritten notes. Use speech-to-text to transcribe and save the speech from media files. You can even create images using AI, directly within DEVONthink.

Store important legal or financial documents in audit-proof databases and as PDF/A. Text in PDFs without text layer and in images is searchable without OCR, and versioning lets you return to previous versions of a document. Edit a document's table of contents, adjust spacings, and fine-tune the WYSIWYG display for Markdown. Organize your favorites, smart groups, and smart rules into groups. We also added new inspectors, most notably a graph that visualizes document relationships. The rewritten web interface of the Server edition is built on a modern foundation, more powerful yet easier to use.

For automators of any skill level, there are new and improved smart actions, including some utilizing AI, as well as many new AppleScript commands. Batch processing now lets you save and reuse configurations.