API and Integrations

Edited by Lawrence Beckwith on September 5, 2026 at 5:45 PM UTC

Ember exposes HTTP interfaces for applications and agents. The standard Compose base URL is http://127.0.0.1:8080/v1; replace the port if your installation uses another value.

Endpoints

Method Path Purpose
GET /health Service health
GET /status Server status
GET /v1/models Advertised model information
POST /v1/chat/completions OpenAI-compatible chat completions
POST /v1/responses Responses-compatible requests
POST /v1/messages Anthropic-compatible messages
POST /v1/completions Legacy completions

Compatibility applies to the features implemented by the installed Ember release. Use its client compatibility documentation when configuring a particular client.

Chat request

curl --fail --show-error http://127.0.0.1:8080/v1/chat/completions \
  -H 'Content-Type: application/json' \
  -d '{
    "model": "deepseek-v4-flash",
    "messages": [{"role": "user", "content": "Hello"}],
    "stream": false
  }'

The default deployment advertises deepseek-v4-flash. Query /v1/models to confirm the name on your instance. A model alias does not load a second model or change its capabilities.

Images

Vision requests use image content parts and require a compatible vision model and projector. Supported image encodings and limits depend on the release. Follow the image request example and image format reference.

Authentication and remote clients

Ember has no built-in authentication. Supplying an API key in a client does not secure the endpoint. Keep the service on loopback, or place it behind an authenticating gateway with restricted network access. Configure any gateway credentials in the client according to that gateway's requirements.

Agent integration

Hermes Agent connects through an OpenAI-compatible custom provider and handles tools, memory, and channel delivery. See Connect Hermes Agent. Desktop control belongs to the agent's KVM integration; Ember does not expose keyboard or mouse endpoints.