AI Providers

HyperVibes uses one shared OpenCode service as its agent execution backend. Every AI provider supported by OpenCode is supported by HyperVibes. See Models.dev for provider and model information. OpenCode runs the LLM and tool calls inside each generated agent workspace.

Connecting a Provider

Manage API-key and OAuth provider connections from the Provider connections page. Provider credentials are stored by OpenCode in its retained data volume, not on individual agents and not in the HyperVibes database.

Provider configuration is global to the shared OpenCode service. Once a provider is connected, select an available model for each scheduled job or conversation. Some models also provide a selectable thinking mode.

AI providers may authenticate with API keys, OAuth, or both. Connect and verify providers from the Provider connections page.

After a provider is connected, disconnected, or completes OAuth, HyperVibes queues a provider configuration reload. The reload waits until OpenCode sessions are idle because it can interrupt in-flight inference. Once idle, HyperVibes refreshes the shared OpenCode provider cache from retained auth storage; existing sessions remain, but an in-flight inference may be interrupted.