Skip to main content
Build and deploy your first AI inference endpoint. Create an account, describe what you need in plain English, run optimization, deploy to a live URL, and make your first API call. About five minutes, start to finish.

Free to start

$10 in free credits on signup. No credit card required.

OpenAI-compatible

Works with OpenAI Python/JavaScript SDKs plus documented LangChain and LlamaIndex examples.

Under 2s cold start

Cached weights keep even scale-to-zero endpoints fast.
Your 10infreecreditsletyoubuild,optimize,andtestpipelinesintheplayground.DeployingtoaliveendpointrequiresapaidCoreplan(customamountfrom10 in free credits let you build, optimize, and test pipelines in the playground. Deploying to a live endpoint requires a paid Core plan (custom amount from 50/mo).
1

Create an account

Sign up at runinfra.ai/sign-up using GitHub or Google. No credit card is required to get started.
2

Describe your pipeline

Open the dashboard and type what you need in plain English:
I need a fast chatbot using Llama 3.1 8B optimized for low latency
The agent builds your pipeline, selects the model, and configures everything automatically. To make changes, just continue the conversation:
Add a response cache and switch to Qwen 2.5 7B instead
3

Optimize

The agent benchmarks your model across GPUs, searches for optimized variants (AWQ, GPTQ, FP8), and identifies the best configuration. You see real-time progress as experiments complete.You can set specific targets before optimization starts:
Optimize for latency, keep cost under $0.10 per request
Optimization takes 2-5 minutes. When it finishes, review the results and select the configuration you want to deploy.
4

Deploy your endpoint

Click Deploy in the deploy tab. RunInfra provisions a GPU endpoint with scale-to-zero and fast cold starts (under 2 seconds). Your endpoint URL and API key appear as soon as the endpoint is ready.
Deployment requires a paid Core plan (custom amount from 50/mo).Your50/mo). Your 10 in free credits let you build, optimize, and test in the playground without upgrading.
5

Use your endpoint

Your endpoint is OpenAI-compatible. Use your RunInfra API key and the RunInfra base URL with OpenAI Python or JavaScript SDKs, or direct HTTP requests.
import os
from openai import OpenAI

client = OpenAI(
    base_url="https://api.runinfra.ai/v1",
    api_key="YOUR_RUNINFRA_API_KEY",
)

response = client.chat.completions.create(
    model=os.environ["RUNINFRA_MODEL"],
    messages=[{"role": "user", "content": "What is RunInfra?"}],
)

print(response.choices[0].message.content)

import OpenAI from "openai";

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

const response = await client.chat.completions.create({
  model: process.env.RUNINFRA_MODEL,
  messages: [{ role: "user", content: "What is RunInfra?" }],
});

console.log(response.choices[0].message.content);

curl https://api.runinfra.ai/v1/chat/completions \
  -H "Authorization: Bearer YOUR_RUNINFRA_API_KEY" \
  -H "Content-Type: application/json" \
  -d "{\"model\":\"${RUNINFRA_MODEL:?Set RUNINFRA_MODEL from GET /v1/models}\",\"messages\":[{\"role\":\"user\",\"content\":\"What is RunInfra?\"}]}"
Replace YOUR_RUNINFRA_API_KEY with your actual API key from Settings > API Keys, and set RUNINFRA_MODEL to a model id returned by GET /v1/models.

​
Next steps

Use cases

Six pre-built workflows: voice, assistants, embeddings, RAG, document AI, transcription.

Prompting best practices

Write better prompts and get better pipelines from the agent.

Deployment

Flex vs Active endpoints, scaling, and cold-start configuration.

API reference

Supported OpenAI-compatible endpoints and fields.