This update is about your first minutes with ToRun: your first song on the Free plan is a gift, a chat you start as a guest in the app comes with you when you sign up, and when a job is out of the Free plan's reach, the message says so in plain words.
Your first song is on us
Making music costs more than the Free allowance covers, so a first try on the Free plan used to be turned down. Now your first song is on us, up to two minutes long, in the app and on the web. In the web music studio, a note tells you before you press Generate that the gift covers it. A longer song needs a balance or a subscription, and the message says so. The web video studio now checks the welcome video gift again after each video, so it never offers a gift you have already used.
Your guest chat comes with you
In the ToRun app you can try ToRun before you have an account. When you then create one, the conversation you had as a guest becomes a chat in your new account, and the app opens it right after your choice on the "Your data & AI providers" screen. The welcome screen is clearer too: it opens with the message box and how many messages you have, and lists what needs an account below it. When the app cannot reach ToRun, it says so, instead of telling you to turn off a VPN or an ad blocker. Signing up is shorter, with the invite code under its own question.
Plain words when the Free plan cannot cover a job
Some jobs cost more than the Free plan can cover, even with full energy. The message now says so instead of suggesting you wait, and offers a one-off Runs pack first, then a paid plan. Deep Research refusals explain themselves the same way. In the web image, video and music studios, Generate stays off after a refusal for settings that cost as much or more, so you can choose a lighter one instead of being turned down again.
Also in this update
- During a voice call, a picture you choose from your Library is now shared with the AI instead of being turned down as invalid.
- When you send a photo with Auto, ToRun answers with a model picked for photo questions on every plan, and if that model fails before it starts, the retry stays on a model picked for photos.
- For meeting customers, the AI interpreter no longer stops speaking a language that falls behind: it speaks a little faster to catch up, never cuts off a sentence it has started, and no longer loses the first words of a sentence. The speaker's own voice plays softly under the translation.
- In a meeting, the video grid fits every face on the screen, the call bar fits every width from a phone to a wide monitor, and the host has a menu on each person's tile to mute them, ask them to turn on their microphone or camera, or remove them.