Model Verification

AIone Model Sources

All models on AIone are accessed through official cloud services to ensure you are using authentic models:

How to Verify Model Authenticity

We do not recommend relying on non-standard response headers as proof of authenticity. A more reliable approach is to check whether the model and usage fields in the response are consistent with your console records.

1. Check the model Field in the Response

Every API response includes the actual model identifier:

{
  "model": "claude-sonnet-4-20250514",
  "choices": [...]
}

The returned model should match the model you requested, or match the current model ID declared in the platform documentation.

2. Verify the usage Field

A normal response typically includes usage.prompt_tokens, usage.completion_tokens, and usage.total_tokens. These values should be proportional to the size of your request and the length of the response.

3. Cross-Reference Console Usage Records

After making calls through AIone, we recommend verifying that the following are consistent:

• The model field in the response
• The usage field in the response
• The model, timestamp, and token consumption shown in the console usage details
• The corresponding billing or usage statistics

4. Retain the Request ID for Troubleshooting

If you have concerns about a specific request, save the id field from the response along with the original request and response content. This makes it easy for the platform to investigate.

AIone's Assurance Measures

1. 100% official channels: No reverse-engineered APIs, unauthorized proxies, or misrepresented models
2. Request transparency: The console shows the model, token consumption, and latency for every call
3. Audit trail: The platform retains complete call metadata for investigating anomalous requests
4. Public status page: portal.aiin1.ai/status displays real-time service status

How to Report a Suspected Issue

If you observe abnormal model behavior or suspect a model authenticity problem:

1. Record the request ID (the id field in the response)
2. Save the complete request and response content
3. Submit a ticket through the Support System -- we will investigate and respond within 24 hours