DoublewordDoubleword
Get started

Dedicated Realtime

Dedicated Realtime gives you an inference deployment with capacity reserved for your traffic, ready for production. We use details like your models, traffic shape and latency targets to tune the infrastructure for the best combination of throughput, latency and price. We tune the deployment with custom kernels and speculators, and can match quantisation to your requirements, so you get the best cost and performance available on the market. Unlike other providers, you still get charged on a per-token basis.

Talk to us about your workload and we will get you started.

What you get

  • Guaranteed performance and SLAs. Your traffic runs on capacity reserved for it, so throughput and latency stay predictable under load.
  • Automatic tuning for your workload. We optimise for the shape of your workload rather than a general case, covering throughput, latency, cache read rate and price together.
  • Region placement. Deployments can run in Europe or the US. Tell us which region you need when you get in touch.
  • A dedicated Slack channel. Talk directly to the engineers running your deployment, for test runs, tuning questions and support.

Calling your endpoint

A dedicated deployment serves the same OpenAI-compatible API on the same base URL. We give you a provider name specific to your org, and your models are addressed with that name in front of them. Swap the model and the rest of your code is unchanged. We confirm your exact model names when the deployment is set up.

from openai import OpenAI

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

response = client.chat.completions.create(
    model="yourorg-zai/GLM-5.3",
    messages=[
        {"role": "system", "content": "You are a helpful assistant."},
        {"role": "user", "content": "What is dedicated 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: 'yourorg-zai/GLM-5.3',
  messages: [
    { role: 'system', content: 'You are a helpful assistant.' },
    { role: 'user', content: 'What is dedicated 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": "yourorg-zai/GLM-5.3",
    "messages": [
      {"role": "system", "content": "You are a helpful assistant."},
      {"role": "user", "content": "What is dedicated inference?"}
    ]
  }'

The Anthropic-compatible Messages API works the same way. See Anthropic API compatibility.

Open Responses API

The Open Responses API gives you a unified interface. Use service_tier: "priority" for realtime.

from openai import OpenAI

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

resp = client.responses.create(
    model="yourorg-zai/GLM-5.3",
    input="Explain quantum computing in one paragraph.",
    service_tier="priority",
)

print(resp.output_text)

Next Steps