Colt Albrecht
  • The /v1/workflows/run endpoint is fire-and-forget — it triggers a run but doesn't return model output in the response, so it can't be used by tools like Continue.dev that need real-time or streamed responses.
  • Continue.dev requires a standard OpenAI-compatible /v1/chat/completions endpoint that returns the model's response directly (ideally with streaming support).
Here's why I think this should already be possible on your end:
Your own API docs at https://api-docs.hatz.ai/home/chat-completions document a /v1/chat/completions endpoint with a curl example showing standard OpenAI-format requests and responses. Your changelog also specifically mentions: "Hatz now offers an OpenAI-compatible API endpoint. If a product supports a custom OpenAI base URL (like many coding tools/agents), customers can point it at Hatz."
That's exactly what I'm trying to do — point Continue.dev (a coding tool) at Hatz.
Could you enable the /v1/chat/completions endpoint for API keys? Specifically:
  1. The endpoint to accept new or existing API keys (via X-API-Key or Bearer)
  2. Standard OpenAI chat completions request/response format
  3. Streaming support (stream: true) if available
  4. Access to Hatz LLM models