OpenAI & Anthropic Compatible API - Eliza

Eliza exposes compatibility endpoints that mirror the OpenAI and Anthropic API formats. Any client, library, or tool built for those APIs (e.g. openai Python SDK, curl scripts targeting /v1/chat/completions) can point at the Eliza server and get agent responses. All paths are at the root (/v1/...) without the /api/ prefix.

Endpoints

Method	Path	Description
GET	`/v1/models`	List available models (OpenAI format)
GET	`/v1/models/:id`	Get a single model (OpenAI format)
POST	`/v1/chat/completions`	Chat completion (OpenAI format)
POST	`/v1/messages`	Create a message (Anthropic format)

GET /v1/models

List available models in OpenAI's /v1/models format. Returns the agent's name and "eliza" as model IDs.

Response

json

{
  "object": "list",
  "data": [
    {
      "id": "eliza",
      "object": "model",
      "created": 1718000000,
      "owned_by": "eliza"
    },
    {
      "id": "Aurora",
      "object": "model",
      "created": 1718000000,
      "owned_by": "eliza"
    }
  ]
}

GET /v1/models/:id

Get details for a single model.

Response

json

{
  "id": "eliza",
  "object": "model",
  "created": 1718000000,
  "owned_by": "eliza"
}

POST /v1/chat/completions

OpenAI-compatible chat completions endpoint. Supports both streaming (SSE) and non-streaming modes. The messages array is processed to extract system and user content, which is sent to the agent for a response.

Request Body

Field	Type	Required	Description
`messages`	array	Yes	Array of `{ role, content }` message objects
`model`	string	No	Model name (used in response metadata)
`stream`	boolean	No	Enable SSE streaming
`room_id`	string	No	Custom room key for conversation isolation

json

{
  "messages": [
    { "role": "system", "content": "You are a helpful agent." },
    { "role": "user", "content": "What is the meaning of life?" }
  ],
  "model": "eliza",
  "stream": false
}

Non-Streaming Response

json

{
  "id": "chatcmpl-uuid",
  "object": "chat.completion",
  "created": 1718000000,
  "model": "eliza",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "The meaning of life is..."
      },
      "finish_reason": "stop"
    }
  ]
}

Streaming Response

When stream: true or the Accept header includes text/event-stream, the response uses Server-Sent Events:

data: {"id":"chatcmpl-uuid","object":"chat.completion.chunk","created":1718000000,"model":"eliza","choices":[{"index":0,"delta":{"role":"assistant"},"finish_reason":null}]}

data: {"id":"chatcmpl-uuid","object":"chat.completion.chunk","created":1718000000,"model":"eliza","choices":[{"index":0,"delta":{"content":"The meaning"},"finish_reason":null}]}

data: {"id":"chatcmpl-uuid","object":"chat.completion.chunk","created":1718000000,"model":"eliza","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

data: [DONE]

Errors

Status	Condition
400	`messages` array is missing or contains no user message
400	Request body contains blocked object keys
503	Agent is not running

POST /v1/messages

Anthropic-compatible Messages API endpoint. Processes the messages array in Anthropic format and returns the agent's response.

Request Body

Field	Type	Required	Description
`messages`	array	Yes	Array of Anthropic-format message objects
`model`	string	No	Model name (used in response metadata)
`system`	string	No	System prompt
`stream`	boolean	No	Enable SSE streaming
`max_tokens`	number	No	Max tokens (accepted for compatibility, not enforced)

json

{
  "model": "eliza",
  "system": "You are a helpful agent.",
  "messages": [
    {
      "role": "user",
      "content": "What is the meaning of life?"
    }
  ],
  "max_tokens": 1024
}

Non-Streaming Response

json

{
  "id": "msg_uuid",
  "type": "message",
  "role": "assistant",
  "content": [
    {
      "type": "text",
      "text": "The meaning of life is..."
    }
  ],
  "model": "eliza",
  "stop_reason": "end_turn"
}

Streaming Response

When stream: true, the response uses Anthropic-style SSE events:

event: message_start
data: {"type":"message_start","message":{"id":"msg_uuid","type":"message","role":"assistant","content":[],"model":"eliza"}}

event: content_block_start
data: {"type":"content_block_start","index":0,"content_block":{"type":"text","text":""}}

event: content_block_delta
data: {"type":"content_block_delta","index":0,"delta":{"type":"text_delta","text":"The meaning"}}

event: content_block_stop
data: {"type":"content_block_stop","index":0}

event: message_delta
data: {"type":"message_delta","delta":{"stop_reason":"end_turn"}}

event: message_stop
data: {"type":"message_stop"}

Errors

Status	Condition
400	`messages` array is missing or contains no user message
400	Request body contains blocked object keys
503	Agent is not running

Usage with Standard SDKs

OpenAI Python SDK

python

from openai import OpenAI

client = OpenAI(
    base_url="http://localhost:2138/v1",
    api_key="your-eliza-api-token"  # or "not-needed" if no token is set
)

response = client.chat.completions.create(
    model="eliza",
    messages=[{"role": "user", "content": "Hello!"}]
)
print(response.choices[0].message.content)

Anthropic Python SDK

python

import anthropic

client = anthropic.Anthropic(
    base_url="http://localhost:2138",
    api_key="your-eliza-api-token"
)

message = client.messages.create(
    model="eliza",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello!"}]
)
print(message.content[0].text)

curl

bash

curl http://localhost:2138/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer your-eliza-api-token" \
  -d '{
    "model": "eliza",
    "messages": [{"role": "user", "content": "Hello!"}]
  }'