Changelog

What's new on ToRun

No marketing copy, just the diff — newest first.

Announcements blog

July 2026

A ticket filed through the contact form or by the assistant was not linked to the person who filed it, so it never appeared on your own Support screen even though our team could see it. New tickets are stamped with their… Read more

Music generation now works straight from the chat and the voice assistant: ask for a song — sung with your lyrics or instrumental — and the track is composed in the background and lands in your Library. We also fixed a p… Read more

A new screen in mobile Settings lists the consents your account has given and when each was given. You can withdraw any of them on the spot, in all 29 languages. Read more

The mobile app is officially out: chat with every major model, generate images, video and music, talk to the realtime voice assistant, and track every Run you spend — all from your phone. Download it from Google Play; th… Read more

On mobile, generated music, speech and videos now keep playing when you switch apps or lock the screen — no more silence the moment the app leaves the foreground. Read more

A full audit closed the gaps where a video render burst, a workflow run, an assistant tool loop or an embedding call could slip past your budget. Every AI spend path now checks the same limits before it runs — so a cap m… Read more

The web voice surface now shows the same funding checks, live cost in Runs, a clear speaking indicator, and friendlier handling of harmless connection blips — the same experience whether you talk to ToRun on your phone o… Read more

Tap the orb on Home and you are in a live voice conversation — the orb even reacts to your microphone level. Your model choice is remembered, voice has its own entry under Tools, and torun://voice deep links straight int… Read more

When a long answer hit the model's output limit, pressing Continue failed with an error. That is fixed on the web — and the mobile app gains the same Continue button, so a cut-off answer can be resumed anywhere. Read more

Running two research jobs side by side and hopping between chats no longer leaves a blank thread or a stuck spinner: switching chats detaches cleanly, reopening reconnects to the live run, and images inside chats load re… Read more

Realtime voice conversations now reach the same tools as text chat: your memory, web search, maps, and image, video and music generation. Ask out loud; it searches, remembers and creates. Read more

Every output of a workflow run — text, images, audio, files — can now be downloaded together as a single ZIP archive instead of saving artifacts one by one. Read more