From Anthropic SDK to OpenAI SDK: Migrate to NovAI Claude in 3 Lines

TL;DR — NovAI's Claude gateway is fully OpenAI-compatible. Python, Node.js, Go, LangChain, LlamaIndex, Dify, Cursor, Continue.dev — they all work. You change one line (the base URL) and one line (the API key). Model names map 1:1 to Anthropic's naming: claude-opus-4-7, claude-sonnet-4-6, claude-haiku-4-5.

Why migrate?

If you are running production workloads on Anthropic's direct API, you are paying:

• $15 / 1M input tokens for Opus 4.7 (NovAI: $8 — save 47%)
• $3 / 1M input tokens for Sonnet 4.6 (NovAI: $2 — save 33%)
• $1 / 1M input tokens for Haiku 4.5 (NovAI: $0.50 — save 50%)

And that's before you factor in the fact that most Chinese developers can't access Anthropic directly without a proxy — NovAI handles the upstream routing for you, with a China-mainland-friendly endpoint.

Python — the 3-line migration

Before (Anthropic SDK)

from anthropic import Anthropic

client = Anthropic(api_key="sk-ant-...")

resp = client.messages.create(
    model="claude-opus-4-5",
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.content[0].text)

After (OpenAI SDK → NovAI)

from openai import OpenAI

client = OpenAI(
+   api_key="sk-novai-...",          # your NovAI key
+   base_url="https://aiapi-pro.com/v1",  # that's it
)

resp = client.chat.completions.create(
+   model="claude-opus-4-7",          # use the 4.7 naming
    max_tokens=1024,
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)

That's it. Three lines changed. No new SDK, no schema rewrite, no tool-calling translation. If you were already using OpenAI's SDK for GPT models, you literally just swap the base_url per request.

Node.js / TypeScript

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.NOVAI_API_KEY,
  baseURL: "https://aiapi-pro.com/v1",
});

const resp = await client.chat.completions.create({
  model: "claude-sonnet-4-6",
  messages: [{ role: "user", content: "Summarize this PR..." }],
  max_tokens: 2048,
});

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

LangChain

from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    model="claude-opus-4-7",
    api_key="sk-novai-...",
    base_url="https://aiapi-pro.com/v1",
    temperature=0.3,
)

result = llm.invoke("Write a SQL query that...")
print(result.content)

Dify / Cursor / Continue.dev

All three tools accept a custom "OpenAI-compatible" provider. Fill in:

Feature support matrix

Environment variable setup

# .env
NOVAI_API_KEY=sk-novai-xxxxxxxxxxxxxxxxxx
NOVAI_BASE_URL=https://aiapi-pro.com/v1

# Python
import os
from openai import OpenAI
client = OpenAI(
    api_key=os.getenv("NOVAI_API_KEY"),
    base_url=os.getenv("NOVAI_BASE_URL"),
)

Rollback plan

Because the only real change is the base URL and the API key, rolling back is trivial. Wrap the client factory:

def make_client():
    if os.getenv("USE_NOVAI") == "1":
        return OpenAI(api_key=os.getenv("NOVAI_API_KEY"),
                      base_url="https://aiapi-pro.com/v1")
    return OpenAI(api_key=os.getenv("OPENAI_API_KEY"))

# Flip one env var to switch providers per deployment.

Further reading

• We Found Claude at 65% Off in China — And Proved It Actually Works
• Claude Opus 4.7 Review: 1M Context, $8/1M Input
• Claude vs GPT-5 vs DeepSeek-V4: Complete 2026 Benchmark
• Browse all 80+ models on NovAI