Providers & Models
Wingman supports multiple model providers and a dedicated voice provider. Models are referenced with the format:
Examples:
anthropic:claude-sonnet-4-5openai:gpt-4ocodex:codex-mini-latestopenrouter:openai/gpt-4ocopilot:gpt-4oxai:grok-betaxai:grok-imagine-image
Configure a provider
Use the CLI to store credentials:
Or supply env vars (see table below). The CLI reads env vars first, then ~/.wingman/credentials.json.
Supported providers
For xai:grok-imagine-image, Wingman uses xAI's native images endpoint (/v1/images/generations) so image output works even when chat-model adapters do not support that model.
Model overrides per agent
You can override the model in agent.json or agent.md:
Provider status
Check which providers are configured:
