Skip to main content
The most common questions when using Global AI. If you cannot find an answer, please contact us.

Accounts and sign-in

First check your spam folder, then confirm you entered the correct email address. If it is still missing, wait a few minutes and try again. If the failure persists, contact support to confirm the platform’s email service is working.
Contact us for manual assistance. To avoid this situation, we recommend binding a third-party account in advance as a backup sign-in method.
If you already registered with an email, third-party sign-in creates a new account by default. Sign in to the original email account first, then go to Personal settings and bind the third-party account. After that, either method signs you into the same account.

Tokens and API calls

No. If lost, go to Token management, delete the token, create a new one, and update your application configuration.
Common causes: the token was copied incompletely, deleted, expired, ran out of quota, or triggered an IP allowlist or model restriction. Go to Token management and check the token status and restrictions.
The model ID is misspelled, or the model is not allowed in your group or token. Search for the model on the Model marketplace to confirm the ID, or adjust the token’s model restrictions.
The platform enforces concurrency and rate limits per user and per token. Reduce concurrency, or add exponential backoff on the client side. If you need a higher limit, contact us.
Confirm that stream=True is set (Python) or the SSE headers are present (curl), and that your client supports chunked reading. Some models do not support streaming. See the Model marketplace for per-model details.

Billing and top-up

Check the order status on the Wallet page. If the order shows success but the balance did not increase, take a screenshot of the successful payment and contact us.
Confirm the code has no extra spaces, has not expired, and was not used by another account. Each code is single-use and cannot be redeemed twice.
Actual quota consumption depends on the input and output token counts. Longer conversation history or longer model responses raise costs. Check per-call token details in the Usage logs.
View the last 30 days of consumption on the Dashboard, then combine it with each model’s multiplier (see the Model marketplace) to estimate future usage.

Client integrations

Use the API base URL shown on the platform home page, typically https://globalai.vip/v1. The native Claude format and some clients require the host URL without /v1 (https://globalai.vip). Follow the client’s instructions. See the Client integration overview for details.
The most common cause is an enabled proxy, VPN, or circumvention tool. Try disabling all proxies and global proxy software, then retry.
Switch to the backup domain https://hk.globalai.vip. Sign-in, top-up, and API calls all work the same way.
Cursor’s third-party API only allows Ask mode and cannot invoke Agent. If you need Agent, subscribe to Cursor’s official plan, or use Claude Code or Cline / Roo Code inside Cursor.
Trae does not support third-party APIs, so you cannot enter a Global AI key directly into its model configuration. Install Claude Code and use its UI extension inside Trae.
Claude Code resends historical tokens every round, and long conversation contexts amplify this. Go to Usage logs, filter by token name, and inspect each request’s token count and actual charge. See the cost notes in the integration overview.
Yes, but we recommend one token per client so you can distinguish usage in Usage logs and minimize the impact if a single token leaks.

Error code cheat sheet

Security and compliance

Immediately go to API keys and delete the token. Check the recent source IPs in the logs. After deletion, create a new token and update every application that depends on it.
Strongly recommended. When your account holds a balance or powers production applications, 2FA significantly reduces the risk of account theft. See Enable two-factor authentication.
The platform only records call metadata (time, model, token count, status) for billing and troubleshooting. Request bodies and response content are not recorded.
Global AI has explicit regional restrictions and prohibited uses (violence, sexual content, politically sensitive material, criminal activity, malicious abuse, and more). Read the full Usage policy before using the platform.