OpenAI-compatible API
Aquilifer's native pass-through to an OpenAI-compatible chat-completions endpoint, for sites that need features the generic API doesn't cover, against OpenAI itself or a self-hosted/local backend.
Sending a message
js
const response = await window.aquilifer.openaiChatCompletions({
messages: [{ role: 'user', content: 'Hello!' }],
});The body matches the real OpenAI chat-completions request shape (minus model, which Aquilifer always sets to whatever the user's bound provider actually is) and returns the real response shape.
Streaming
A dedicated streaming method, openaiChatCompletionsStream(body), yields the provider's own raw streaming events, not a simplified shape.
If the user isn't bound to an OpenAI-compatible provider
Calling this when the user's bound provider doesn't match its type doesn't just fail: it prompts the user to switch (see Connecting a site). Design for that as a normal, expected outcome, not an error path to work around.