A new detection model, trained on more real mail. - On our test mail it flags about a third fewer legitimate emails and catches more than twice as much spam as the previous model, while using a quarter less memory. - It now catches the fake "your Prime trial is ending" notices that open the fake-bank-adviser scam, in French as well as English. - Spam that reaches your Inbox never shows up on the dashboard, so the dashboard now says so and tells you how to forward one to us. - A small card asks whether you enjoy Klar. Yes takes you to the App Store; not really takes you to us. - Clearer German and French labels.

Klar now shows you its work before it moves anything. - For your first 30 days, spam is flagged in red instead of being moved, so you can check what Klar catches before you trust it. After that it moves to Junk on its own, and you can pick either behaviour whenever you like. - The dashboard counts the mail Klar has been through, lists what it caught, and keeps the rest of the history behind Show more. - Got one wrong? Report several catches at once from the dashboard, with the original message attached so we can fix it. - Klar explains why it needs your Contacts before macOS asks, and asks only once. - The Aggressiveness setting now changes what the filter does. It was showing three choices and applying the most careful one whatever you picked. - Clearer German and French throughout the app.

Fewer false alarms on mail you actually want. - Order confirmations, delivery updates and payment receipts from brands that authenticate their mail now stay in your inbox. On our test set, wrong flags on this kind of mail dropped by about 60%. - Replies from real people are no longer flagged because of common words in the sender name or subject.