Domain Migration Guide

LinkCompute's API domain has switched from ai.futurecore.com.cn to api.futurecore.com.cn.

There's only one change: the prefix goes from ai. to api.. Everything else stays the same — paths, protocols, the /v1 rule, your API Key, token groups, billing prices, and account balance.

At a glance

Change only the hostname ai → api; leave everything after it untouched.

👋 Non-technical user (Cherry Studio and other desktop clients)

You don't need to know what a "domain" is — just change that one address line in your client and leave the key alone.

Cherry Studio steps

1. Open Cherry Studio and go to Settings → Model Providers
2. Find the LinkCompute provider you set up earlier (an OpenAI-compatible provider)
3. In the Base URL field, change ai.futurecore.com.cn to api.futurecore.com.cn
   • Before: https://ai.futurecore.com.cn
   • After: https://api.futurecore.com.cn
4. Leave the API Key (sk-...) untouched and save
5. Go back to the chat window, send a quick hello — a normal reply means you're done

Other desktop clients

Same logic everywhere: find the field where you entered the Base URL / API address, swap ai. for api., leave the key as is, save, and send a message to verify.

👨‍💻 Software engineer (SDKs / your own services)

Swap the Base URL in your code and config for the new domain. The API Key, request bodies, and call patterns don't change at all.

Where to change it

OpenAI SDK (Python)

from openai import OpenAI

client = OpenAI(
    api_key="sk-xxxx",
    base_url="https://api.futurecore.com.cn/v1",  # old: https://ai.futurecore.com.cn/v1
)

OpenAI SDK (Node.js)

import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: process.env.OPENAI_API_KEY,
  baseURL: 'https://api.futurecore.com.cn/v1', // old: https://ai.futurecore.com.cn/v1
});

Environment variable

# old: export OPENAI_BASE_URL="https://ai.futurecore.com.cn/v1"
export OPENAI_BASE_URL="https://api.futurecore.com.cn/v1"

curl sanity check (do this first after migrating, to confirm the path works)

curl https://api.futurecore.com.cn/v1/chat/completions \
  -H "Authorization: Bearer sk-xxxx" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-5.4","messages":[{"role":"user","content":"hello"}]}'

Codebase-wide checklist

The old domain may be hiding in more than one place. Grep for ai.futurecore.com.cn everywhere and replace it with api.futurecore.com.cn:

• Hardcoded Base URLs in source
• .env / .env.local / .env.production and other env files
• Secret managers / config centers (Vault, Nacos, Consul, cloud Secrets Managers)
• CI/CD pipeline variables and build args
• Config baked into container images, ENV in Dockerfile, k8s ConfigMaps / Secrets
• Reverse proxy / gateway upstream configs
• Hardcoded addresses in client apps and mini-programs

# From the project root, sweep to confirm nothing is left behind
grep -rn "ai\.futurecore\.com\.cn" .

Verify

After switching, go to Usage Logs, confirm by Request ID or token name that new requests are billed correctly, then retire the old config.

🤖 Heavy AI developer (Agent / Claude Code / Codex CLI)

You likely have multiple tools, tokens, and config locations running at once. Since the old domain is already retired, you need to switch everything promptly — but do it tool by tool and verify each one, so if a tool breaks you can pinpoint it immediately instead of flipping everything and having nowhere to look.

Where to swap the domain

1. CC Switch: edit your LinkCompute configuration, change the API Base URL to https://api.futurecore.com.cn, save, and switch to that config
2. Claude Code: on the Anthropic-compatible path, set ANTHROPIC_BASE_URL (or the matching config) to https://api.futurecore.com.cn; on the OpenAI-compatible path, use https://api.futurecore.com.cn/v1 with the /v1 suffix
3. Codex CLI: change the base_url / env var in its config file to the new domain
4. Cursor / Cline and other editor plugins: change the Base URL in each tool's custom OpenAI-compatible provider settings
5. Agents / automation pipelines: check every server-side env var, secret store, scheduled job, container image, and queue consumer for hardcoded old domains

Smooth-cutover tips

• This is where one-token-per-tool pays off: switch one tool at a time, then immediately check Usage Logs by token name to confirm it's hitting the new domain and that fresh requests are coming in, before moving to the next
• Task-type calls (async tasks from MJ, Suno, Kling, Seedance) — after switching, confirm in Task Logs that tasks submit and call back normally
• Watch your balance: on migration day, keep an eye on Billing and usage statistics to confirm call volume and spend stay normal; if a tool suddenly stops sending new requests, its address probably isn't fully updated
• Retries and backoff: if your Agent has automatic retries, make sure failures against the old domain don't retry endlessly and flood your logs — fix the address first

Post-migration error cheat sheet

Still unresolved? See the FAQ or reach out via Community & Support.