DoublewordDoubleword
Get started

Realtime Inference

The self-serve realtime API is designed for development and testing, letting you quickly evaluate models, iterate on prompts and prototype your pipeline. It runs on shared, rate-limited infrastructure.

For production realtime workloads, talk to us. We’ll review your workload and configure an optimised deployment around your requirements, giving you dedicated capacity and the best combination of throughput, latency, caching and price.

For production workloads that don't need instant responses, consider Async Inference or Batch Inference for significant cost savings.

Quick Start

Using the Playground

The fastest way to test the realtime API is through our interactive playground. Simply select a model, enter your prompt, and get instant responses.

Chat Completions

from openai import OpenAI

client = OpenAI(
    base_url="https://api.doubleword.ai/v1",
    api_key="{{apiKey}}"
)

response = client.chat.completions.create(
    model="{{selectedModel.id}}",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "What is batch inference?"}
    ]
)

print(response.choices[0].message.content)
import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://api.doubleword.ai/v1',
  apiKey: '{{apiKey}}'
});

const response = await client.chat.completions.create({
  model: '{{selectedModel.id}}',
  messages: [
    { role: 'system', content: 'You are a helpful assistant.' },
    { role: 'user', content: 'What is batch inference?' }
  ]
});

console.log(response.choices[0].message.content);
curl https://api.doubleword.ai/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer {{apiKey}}" \
  -d '{
    "model": "{{selectedModel.id}}",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "What is batch inference?"}
    ]
  }'

Open Responses API

The Open Responses API provides a unified interface with built-in support for background processing. Use service_tier: "priority" for realtime inference:

from openai import OpenAI

client = OpenAI(
    base_url="https://api.doubleword.ai/v1",
    api_key="{{apiKey}}"
)

# Blocking — waits for the response
resp = client.responses.create(
    model="{{selectedModel.id}}",
    input="Explain quantum computing in one paragraph.",
    service_tier="priority",
)

print(resp.output_text)

Background Mode

For longer-running requests, use background=True to return immediately and poll for the result:

from openai import OpenAI
from time import sleep

client = OpenAI(
    base_url="https://api.doubleword.ai/v1",
    api_key="{{apiKey}}"
)

resp = client.responses.create(
    model="{{selectedModel.id}}",
    input="Write a detailed essay about the history of space exploration.",
    service_tier="priority",
    background=True,
)

# Poll until complete
while resp.status in ("queued", "in_progress"):
    print(f"Status: {resp.status}")
    sleep(2)
    resp = client.responses.retrieve(resp.id)

print(f"Done! Output:
{resp.output_text}")

Background mode still runs at priority pricing and starts immediately — it simply lets you poll for the result instead of holding the connection open. To trade latency for a lower rate, use Async Inference with service_tier: "flex".

Next Steps