Models & Providers

Providers control how Gestura hears you, thinks, and speaks back. Most users only need a simple mental model: local is the safe default, cloud is optional when you want it, and fallback settings keep the app usable when your preferred provider is unavailable.

The three provider families

• Speech to text: turns your voice into text.
• Language model: plans, reasons, and generates responses.
• Text to speech: turns responses back into audio.

Local first, cloud when you choose it

Gestura defaults to a local voice provider path. You can configure cloud providers for speech, text, or reasoning when you want different latency, quality, or capability tradeoffs. The important thing is to understand which parts of your workflow remain local and which parts now depend on a remote service.

Why fallback settings matter

Fallback models and provider defaults help Gestura recover gracefully from rate limits, unavailable services, or experimental settings. If you are tuning aggressively, keep one known-good baseline model configured so the app always has a dependable path.

Managing secrets safely

API keys should move into secure storage whenever possible. Keep plain-text config focused on durable structure, not on secrets that belong in the system keychain or temporary environment variables.

How to choose a good setup

• Start with the defaults that already work.
• Change one provider family at a time.
• Validate after each change.
• Use fallback settings so experiments never leave you without a working path.