OpenAI-compatible client - MESHROUTER

MESH speaks the OpenAI Chat Completions, Embeddings, and Models APIs verbatim. Any OpenAI-compatible client (the official OpenAI SDK, LangChain, LlamaIndex, Vercel AI SDK, etc.) works by changing two settings:

Setting	MESH value
Base URL	`https://api.meshrouter.app/v1`
API key	`mesh_live_...`

TypeScript (official `openai` package)

import OpenAI from 'openai';

const client = new OpenAI({
  baseURL: 'https://api.meshrouter.app/v1',
  apiKey: process.env.MESH_API_KEY,
});

const reply = await client.chat.completions.create({
  model: 'auto', // or "mesh/claude-sonnet-4.5"
  messages: [{ role: 'user', content: 'hello' }],
});

Python (official `openai` package)

from openai import OpenAI

client = OpenAI(
    base_url="https://api.meshrouter.app/v1",
    api_key="mesh_live_...",
)

reply = client.chat.completions.create(
    model="auto",
    messages=[{"role": "user", "content": "hello"}],
)

cURL

curl https://api.meshrouter.app/v1/chat/completions \
  -H "Authorization: Bearer $MESH_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "mesh/claude-sonnet-4.5",
    "messages": [{"role":"user","content":"hello"}]
  }'

Receipt + cost metadata still come back on the response headers (x-mesh-receipt-id, x-mesh-total-usd, x-mesh-margin-usd, x-mesh-upstream-cost-usd) regardless of which client you use.

When to use the native MESH SDK instead

The official OpenAI SDKs don’t know about MESH-specific surfaces like payments, account balance, or signed attestation. If you want typed helpers for those, use:

They keep the OpenAI-compatible chat shape and add typed methods for receipts, payments, usage, and account management.