Pipe Network — The AI Cloud

DROP-IN API

Compatible with the SDKs your team already uses

Point your existing OpenAI client at Pipe and route across any provider with one parameter. No SDK migration. No rewrites. No vendor lock-in.

OpenAI, Anthropic, and standard OSS SDKs
Drop-in: change the base URL, keep your code
Set model="auto" for cost-aware routing

          python
          typescript
          curl
        

from openai import OpenAI

client = OpenAI(
  api_key="pipe_sk_...",
  base_url="https://api.pipe.network/v1"
)

# Route across providers with one parameter
response = client.chat.completions.create(
  model="auto",
  messages=[{"role": "user",
              "content": "Summarize Q3 earnings"}],
  routing={"policy": "cost_aware",
           "fallback": "auto"}
)
import OpenAI from "openai";

const client = new OpenAI({
  apiKey: "pipe_sk_...",
  baseURL: "https://api.pipe.network/v1",
});

// Route across providers with one parameter
const response = await client.chat.completions.create({
  model: "auto",
  messages: [{ role: "user",
              content: "Summarize Q3 earnings" }],
  routing: { policy: "cost_aware",
             fallback: "auto" },
});
curl https://api.pipe.network/v1/chat/completions \
  -H "Authorization: Bearer pipe_sk_..." \
  -H "Content-Type: application/json" \
  -d '{
    "model": "auto",
    "messages": [
      {"role": "user", "content": "Summarize Q3 earnings"}
    ],
    "routing": {
      "policy": "cost_aware",
      "fallback": "auto"
    }
  }'
      

WHY PIPE

AI workloads need more than model access

Modern AI applications depend on reliable model access and fast delivery of large data. Teams need to route requests across providers, serve model files and datasets closer to users, reduce egress costs, and keep applications online when providers degrade.

Pipe brings model routing and edge data delivery into one AI Cloud.

Route AI requests

Connect any hosted model, open-source AI service, or private endpoint through one API.

Deliver data

Serve model files, datasets, media, software assets, and S3 objects from a global edge network.

Control cost

Reduce egress, manage usage, route by policy, and optimize infrastructure spend across providers.

FOR PRODUCTION AI TEAMS

Built for AI teams that don't want another internal router project

Most teams start with one model provider. Then they add another. Then fallback logic, usage tracking, cost controls, evals, rate limits, private endpoints, and provider-specific SDKs. Eventually, they're maintaining their own AI infrastructure.

Pipe gives teams a managed routing layer for AI traffic, plus the edge delivery layer needed for AI data — so engineering stays focused on the application, not the plumbing.

USE CASES

Apps. Agents. Platforms

AI applications

Route user requests across providers based on cost, latency, availability, or model quality.

Agent platforms

Keep agents online with fallback routing, provider controls, and request-level observability.

Model platforms

Serve model files, weights, datasets, and AI traffic from one infrastructure layer.

Media AI

Deliver large files, video assets, generated media, and model outputs from the edge.

Enterprise AI

Centralize model access, team usage, budgets, provider policies, and private endpoint routing.

AI Read Layer

Model routing. Edge data delivery