Connecting AI
Vekrun ships no model access of its own. You connect the providers you already use, and you pay them directly. There is no Vekrun account and no Vekrun subscription.
Three ways to connect
Section titled “Three ways to connect”- Your own API key. Anthropic, OpenAI, Google Gemini, or any OpenAI-compatible endpoint. Works today.
- A model on your machine. Vekrun detects Ollama and LM Studio. Works today; see Choosing models.
- An existing coding-agent account. Planned, and only through documented sign-in flows. A consumer chat subscription is not an API key, and Vekrun will not pretend it is.
Add a provider with an API key
Section titled “Add a provider with an API key”- Open Settings, then Providers.
- Choose the provider, name the connection and paste your key.
- Press Test. Vekrun makes a live request and tells you whether the key works.
- Save.
The key is stored in your operating system’s keychain, never in a plain file. Removing the connection deletes the key.
What happens before anything is sent
Section titled “What happens before anything is sent”The first time a session would send something to a hosted provider, Vekrun shows you exactly what is about to leave your machine and waits for you to press Send. See Coding sessions.
Hosted providers bill you at their own rates. Local models cost nothing per request. Vekrun shows usage per session and says “unknown” where a provider does not publish its pricing.