Powering your twin

Sign in on first launch (email or Google) and the brain is connected, that is the whole setup. Your tasks run on Twinly's managed engine, which pairs a strong reasoning model with a vision model for the steps that need eyes, tuned for the way the twin drives real apps. The Free plan gives you 5 tasks a day with every feature included, no card needed, so you can put the twin to work before paying anything.

Subscribe, switch plans, or cancel any time in the app under Account; checkout opens through Stripe in your browser and the plan lands on your account within a few seconds. Cancelling keeps the plan running to the end of the billing period. Extra usage tops up from $5 whenever you want more, and referral credit ($10 to you and $10 to a friend who joins with your code) lands on the same balance.

Each paid plan includes a daily and a weekly allowance, shown as a simple percentage in the app (the sidebar gauge and the Account screen), so you always know where you stand without doing math. Different tasks cost different amounts: a quick text or a calendar change is light, while long vision-heavy computer-use runs are the heaviest. Allowances reset on their own schedule and do not roll over; top-ups and referral credit burn only after the included allowance.

Builds from the original free beta work differently: there is no Twinly account requirement, and the twin thinks with your own AI provider key from Anthropic, OpenAI, Azure OpenAI, or a custom OpenAI-compatible endpoint. You pay that provider directly for usage, and the key lives in your OS keychain (the macOS Keychain, or Windows Credential Manager on a PC), never leaving your device. Those builds keep working; everything below applies to them.

1. Get a key from your provider's dashboard (Anthropic Console, OpenAI Platform, or your Azure resource).
2. In Twinly, open Settings and select the provider.
3. Paste the key and save. The provider chip at the bottom of the app flips to connected.

Azure OpenAI

• Endpoint: your resource URL, like https://your-resource.openai.azure.com/.
• Deployment: the deployment name you created in Azure AI Foundry. This is the model field for Azure.
• Key: from the same resource page.

The brain / vision / call model fields

The twin does three different jobs, and each wants a different kind of model. Beta builds let you set them independently in Settings: Brain (plans tasks, writes drafts, runs the agent loop), Vision (reads screenshots while the twin works your device), and Call (holds live phone conversations, wants low latency). Leave vision or call blank and the brain model covers that job too. Best results come from gpt-5.4-mini and up (OpenAI / Azure) or Claude Opus 4.6 and up; smaller tiers (nano, flash, 4o-mini, 3.5, Haiku-3) tend to hallucinate when driving your real apps.

Roughly what it costs on your own key

The bill is dominated by the screenshots the twin reads while working your device and by long multi-step reasoning runs. Light use (a few texts, calendar, email a day) lands at a few dollars a month; a daily mix of computer-use tasks runs low tens of dollars; heavy vision-driven use scales with the model you pick. Check your provider's usage dashboard for the real number. You can switch providers any time, and live calls fall through to any other configured provider mid-call rather than dropping the conversation.