Image requests that mention documents now generate an image instead of refusing. Auto mode offers the video model chooser. Clearer pricing page layout, plus analytics and stability improvements.

Behind-the-scenes fixes so your account activity is recorded accurately, plus general stability improvements.

Organization owners can now assign tokens to teammates from a single screen — pick a member, choose an amount, and see the allocation history with each recipient named. Tap any history row to pre-fill that member again.

Tapping the mic no longer opens the keyboard over dictation, and the composer buttons stay out of your way. Starter prompts now fill the message box so you can edit them before sending, instead of sending straight away. Exported PDFs and Word files render properly again. On the home screen, tap a day to see which teammate used what. Smarter automatic model routing, and clearer pricing before you generate a video.

Redesigned voice mode with a cleaner interface. A live waveform while dictating, so you can see what the mic hears — and discard a recording without sending it. Documents now read properly on phones, downloads work throughout the app, and Google sign-in is fixed.

Redesigned voice mode with a cleaner, calmer interface. New live waveform while dictating, so you can see exactly what the mic is hearing — and discard a recording without sending it. Larger, easier-to-hit controls, and a fix for titles pushing buttons off-screen.

• Sign in with Google is now available on iPhone • Fixed an issue that could prevent sign-in from completing • Clearer messages when sign-in fails, shown right above the form • Free accounts now include 10 messages per month across 6 models • Refreshed chat experience with a deeper dark theme • Stability and performance improvements

Sign in with Google - now works right in the app. New AI video and avatar tools: create short reels, bring images to life as talking avatars, turn images into video, and reuse custom characters. Newest models: Claude Opus 5, Sonnet 5, Fable 5, and GPT-5.6. Apollo (multi-model council) improvements, plus performance and stability fixes.

New models and a batch of fixes and improvements: - New models: Claude Sonnet 5, Claude Fable 5, Google Veo 3 (video), and more of the latest models across providers. - Read Aloud now shows the token cost before it plays. - Fixed: changing your password and uploading a profile photo. - Fixed: a sign-in issue where login didn't take effect until you reopened the app. - Fixed: images and avatars no longer occasionally disappear. - Voice mode now respects the screen's safe areas; polished chat and model-picker layout. - More accurate video/image generation pricing, plus other bug fixes and performance improvements.

A fresh new look and a big voice fix: - Brand-new app icon and a redesigned Home screen — your token balance, usage, and quick actions at a glance. - New model picker: browse models grouped by provider, with dedicated Image and Video generation, search, and capability tags. - A new welcome screen with quick-start suggestions when you open a chat. - Apollo is now a simple on/off toggle. - Fixed: the AI's voice now plays reliably. - Fixed: changing your password and updating your profile photo. - Performance and stability improvements.

Apollo gets more powerful: compare answers from multiple AI models side by side in a matrix view, expand any cell to read the full response, and re-run a comparison in one tap. Documents and letters now export as clean, premium PDFs with selectable, copyable text. The assistant now knows the current date and time in your time zone, for more accurate and timely answers. More reliable file uploads, plus various bug fixes and performance improvements.