Personal Space Guide

1. Model Plaza

Before officially starting to call models, you can enter the "Model Plaza" to view the list of models currently supported for calling. Over 100 global models have been launched, and more will continue to be tracked and launched in the future.

• Multi-dimensional Filtering: You can quickly locate models by model vendor, capability tags, discount groups, billing types, endpoint types, and other dimensions.
• Price Transparency: Each model card clearly displays the model's [Official Original Price]. Click the details page to view model information, endpoint information, exclusive discount group prices, context length, and other information. Actual calling costs are based on the group bound to the current API Key and the billing rules hit by the call.
• Technical Support: Click API documentation to quickly navigate to technical documentation for calling details. https://neolink.com/docs/index

2. API Key Management

2.1 Creation Process

STEP1: Enter the "API Key" page

STEP2: Click "+ Create API Key" and configure basic information

To successfully create a key, you need to enter Key Name, set Discount Group, and set Expiration Time.

⚠️ Please note: "Group" is required and all are selected by default, which supports a single key calling all models.

• Discount Tied to Business: The platform will dynamically display your exclusive discount rate (e.g., x0.8, x0.9, etc.) in the group dropdown menu based on business agreements.
• Official Price Channel: If you select the "Official Price" group, it means the Key executes the standard pricing of the model vendor with no additional discounts.

STEP3: Other Optional Configurations

If you have more requirements, you can choose Quota Limit, Access Model List, IP Whitelist and other advanced features for refined management.

• Quota Limit: You can set a specific usage amount limit for each API Key, or check "Unlimited Quota". Once the Key consumption reaches the set amount, the system will automatically stop the Key's calling service to prevent budget overruns.
• Model List: You can select specific models that the Key is allowed to call to achieve permission isolation. For example, you can configure a test Key to only call cheaper models while restricting it from calling high-cost models. If not filled, it defaults to supporting all available models under that group.
• IP Whitelist: Supports entering specific IP addresses allowed to call this API. Through IP Whitelist, even if the API Key is accidentally leaked, servers not in the whitelist cannot initiate requests, fundamentally eliminating the risk of unauthorized use. If not filled, there is no restriction on access sources.

2.2 API Key List Management

• Search and Filter: Supports quick retrieval through "Name" or "API Key" strings. When there are many Keys, it helps you find target entries in seconds.
• Quick Management:
  • Status Toggle: Click the "Disable" icon to disable the Key with one click; click "Enable" to restore service. When a Key leak is suspected, this is the fastest protective measure.
  • Edit and Update: Click the "Pencil" icon to modify quota, IP whitelist, or model list and other advanced configurations at any time.
  • Delete Credential: Click the "Trash" icon to permanently remove the Key.
• Data Refresh: Click the "Sync/Refresh" icon in the upper right corner to synchronize the Key's remaining quota status in real-time.

3. Monitoring and Analysis

The platform provides refined statistics functions, making the destination of every penny clearly visible.

3.1 Calling Logs

In "Calling Logs", you can track the details of every request in real-time, including calling time, model used, Token consumption, and specific cost.

• Precise Search and Filtering: The log system supports high-performance retrieval functions. Click the "Expand" button to precisely locate data.

  • Time Filter: Customize query time range.
  • API Key/Group Filter: Lock specific project or user consumption based on specific "API Key Name" or "Group Name".
  • Model Search: Enter model keywords to view the calling frequency and success rate of that specific model.
  • Log Type: Distinguish between recharge, consumption, management, system, error, and other different types of business logs.

• Quick Dashboard: The top of the page displays in real-time the Total Consumption, RPM, and TPM within the current filter range, giving you an overview of performance trends during that period.

  • Consumption Amount: The actual total cost under the current filter conditions.
  • RPM (Requests Per Minute): Number of requests per minute, measuring business concurrency pressure.
  • TPM (Tokens Per Minute): Token consumption per minute, measuring model load.

• Log Field Analysis

Field Name	Description
Duration/First Token	Records the total request response time and the delay of the first token generation, which is a core metric for measuring LLM experience.
Input/Output	Details the number of Input Tokens and Output Tokens consumed by this request.
Cost	The actual deducted amount calculated based on model unit price and Token consumption. You can use this field to precisely calculate the real cost of each conversation.
Details/Status	Click the details icon on the right to view request status. If it shows failed, the details will include error codes.

3.2 Resource Statistics

On the "Resource Statistics" page, you can view summarized visualization charts:

• Core Financial and Usage Metrics: The dashboard top displays real-time statistics of all data within the selected time range.
  • Total Consumption: Shows the actual cost of all API calls under the current filter conditions.
  • Total Requests: Records the total number of successful requests.
  • Total Tokens: Shows the total Token consumption (including input and output), which is the most core basis for billing.
• Key Performance Metrics: Determine backend business concurrency pressure.
  • Average RPM (Requests Per Minute): Average number of requests per minute. If the value spikes abnormally, there may be malicious API calls or business explosion growth.
  • Average TPM (Tokens Per Minute): Average Token consumption per minute, reflecting the text throughput processed per unit time.
• Model Data Analysis and Distribution Charts: Display model calling trends by day.
  • Model Consumption Distribution: Shows the consumption proportion of different models (such as Claude-3, DeepSeek, etc.) in bar charts or pie charts.
  • Model Calling Distribution: Analyzes which models are called most frequently.

4. Recharge and Alerts

4.1 Recharge Methods

Enter the "Balance Recharge" page.

• Redemption Code Recharge: Click "Redemption Code Recharge" and enter the code you obtained.
• Offline Recharge: If you don't have a redemption code, please contact the administrator to manually recharge in the backend.

4.2 Balance Alerts

To prevent service interruption due to balance exhaustion, please be sure to set balance alerts: Go to "Balance Recharge", set the Balance Alert Value, and choose the notification method. When the balance falls below this value, the system will automatically remind you via SMS or Email.