Streaming - Mako

Overview

When stream: true, the gateway streams the response as Server-Sent Events (SSE). Mako’s streaming goes beyond standard OpenAI streaming — it also emits custom events for tool execution, giving your frontend real-time visibility into the agent’s actions.

Event types

`content_delta`

Standard OpenAI-compatible content chunk. Your OpenAI SDK handles these automatically.

{
  "id": "chatcmpl-1718464968543",
  "object": "chat.completion.chunk",
  "model": "mako-32b-conductor",
  "choices": [
    {
      "index": 0,
      "delta": { "content": "Aerodrome currently has " },
      "finish_reason": null
    }
  ]
}

`tool_start`

Emitted when the agent begins executing a tool.

{
  "type": "tool_start",
  "tool": "get_eth_balance",
  "args": { "address": "0xd8dA...", "chain": "base" }
}

`tool_trace`

Emitted when a tool completes. Contains the result or error.

{
  "type": "tool_trace",
  "tool": "get_eth_balance",
  "ok": true,
  "result": {
    "address": "0xd8dA6BF26964aF9D7eEd9e03E53415D37aA96045",
    "balance_eth": "5.688914350696581971",
    "chain": "Base"
  }
}

`agent_text`

Intermediate text from the model during tool rounds (e.g., reasoning about which tool to call next). Not the final answer.

{
  "type": "agent_text",
  "content": "let me check that balance on Base"
}

`done`

Final event. The last two SSE messages are always:

data: {"id":"chatcmpl-...","choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

data: [DONE]

Consuming the stream

JavaScript
Python

const response = await fetch(
  "https://gateway.deepmako.com/v1/chat/completions",
  {
    method: "POST",
    headers: {
      "Content-Type": "application/json",
      "x-wallet-address": "0xYourWalletAddress",
    },
    body: JSON.stringify({
      model: "conductor",
      messages: [{ role: "user", content: "what is gas on base?" }],
      stream: true,
    }),
  }
);

const reader = response.body.getReader();
const decoder = new TextDecoder();

while (true) {
  const { done, value } = await reader.read();
  if (done) break;

  const text = decoder.decode(value);
  for (const line of text.split("\n")) {
    if (!line.startsWith("data: ")) continue;
    const payload = line.slice(6);
    if (payload === "[DONE]") break;

    const event = JSON.parse(payload);

    if (event.type === "tool_start") {
      console.log(`🔧 Calling ${event.tool}...`);
    } else if (event.type === "tool_trace") {
      console.log(`✅ ${event.tool}: ${JSON.stringify(event.result)}`);
    } else if (event.choices?.[0]?.delta?.content) {
      process.stdout.write(event.choices[0].delta.content);
    }
  }
}

from openai import OpenAI

client = OpenAI(
    base_url="https://gateway.deepmako.com/v1",
    api_key="not-needed",
    default_headers={"x-wallet-address": "0xYourWalletAddress"}
)

stream = client.chat.completions.create(
    model="conductor",
    messages=[{"role": "user", "content": "what is gas on base?"}],
    stream=True,
)

for chunk in stream:
    if chunk.choices[0].delta.content:
        print(chunk.choices[0].delta.content, end="")

Model naming

Use the alias conductor in your requests. The response will include the canonical model ID mako-32b-conductor. Both the alias and the canonical ID are accepted in the model field.

​Overview

​Event types

​content_delta

​tool_start

​tool_trace

​agent_text

​done

​Consuming the stream

​Model naming