API keys and authentication
Create server-side keys, rotate exposed secrets and fix 401 errors.
Open guidanceHelp center
Search common questions, follow the right documentation path or collect the details support needs to debug an account-specific issue.
Sign in with Google, open the console and create a key in the API key area. Store the key server-side and rotate it immediately if it is exposed.
Use https://thesmartoken.com/v1 as the base URL. Chat completion requests use the standard /chat/completions path under that base URL.
Choose by workload. Start with DeepSeek for coding or cost-effective reasoning, Kimi for longer context and Qwen for multilingual product behavior, then verify with your own prompts.
Reduce concurrency, add backoff and retry later. If it repeats, include timestamp, endpoint path, model ID and whether the request was streaming when you contact support.
Add balance, shorten prompts, lower max output length or switch to a lower-cost model. Review wallet history if the spend does not match expectations.
No. Keep SmarToken keys in backend code, serverless functions, workers or secure automation environments. Browser apps should call your backend.
Public release notes and maintenance notices appear on the announcements page when an announcement is enabled by the operator.
Include account email, request time, model ID, endpoint path, error code, approximate prompt size and reproduction steps. Do not paste private API keys.
Start here
Create server-side keys, rotate exposed secrets and fix 401 errors.
Open guidanceChoose DeepSeek, Qwen, Kimi or another configured route for a specific workload.
Open guidanceUnderstand token pricing, promotional credit, insufficient balance and payment channels.
Open guidanceUse the OpenAI SDK shape, the SmarToken base URL and streaming responses.
Open guidanceMove China-model traffic with model ID mapping and a controlled test set.
Open guidanceReview budgets, fallback routes, support information and operational checks.
Open guidanceDiagnostics
A ticket is easier to resolve when the initial evidence already separates authentication, route, cost and latency issues.
| Issue | What to check |
|---|---|
| Authentication failed | Check Bearer token format, key status and whether the request is running server-side. |
| Model not found | Confirm the model ID exists in the public catalog and is spelled exactly as documented. |
| High cost | Compare prompt size, output length, retries and selected model price. |
| Slow response | Check streaming mode, model family, prompt size and upstream route availability. |
Common questions
Sign in with Google, open the console and create a key in the API key area. Store the key server-side and rotate it immediately if it is exposed.
Use https://thesmartoken.com/v1 as the base URL. Chat completion requests use the standard /chat/completions path under that base URL.
Choose by workload. Start with DeepSeek for coding or cost-effective reasoning, Kimi for longer context and Qwen for multilingual product behavior, then verify with your own prompts.
Reduce concurrency, add backoff and retry later. If it repeats, include timestamp, endpoint path, model ID and whether the request was streaming when you contact support.
Add balance, shorten prompts, lower max output length or switch to a lower-cost model. Review wallet history if the spend does not match expectations.
No. Keep SmarToken keys in backend code, serverless functions, workers or secure automation environments. Browser apps should call your backend.
Public release notes and maintenance notices appear on the announcements page when an announcement is enabled by the operator.
Include account email, request time, model ID, endpoint path, error code, approximate prompt size and reproduction steps. Do not paste private API keys.
Use support for account-specific billing, key or route issues. Public help pages cannot expose private account data.