Request Logs

ZenMux provides a comprehensive logging system to help you monitor and analyze all API call records in real time. With logging, you can review detailed information for each request—including token usage, cost, performance metrics, and more—so you can better optimize your application and control costs.

Viewing Logs

Logs UI

On the Logs page in the console, you can view detailed records for all API calls. The UI provides powerful filtering and analysis capabilities:

Filters:

• Time range (Time): Select a specific date range to view historical records
• API key (API Keys): Filter logs by different API keys for easier multi-project management
• Request ID search (Request): Enter a request ID to quickly locate a specific request
• Provider (Provider): Filter by provider, such as Anthropic, OpenAI, etc.
• Model (Model): Filter by model to quickly find call records for a specific model
• Finish reason (Finish Reason): Filter by finish reason, such as stop, end_turn, max_tokens, etc.

Log list fields:

Field Name	Description
Timestamp	Timestamp when the request was initiated
Model	Model name used (e.g., openai/gpt-5.2, anthropic/claude-opus-4.5)
Input Tokens	Number of input tokens; click to view a detailed token breakdown
Output Tokens	Number of output tokens
Cost	Cost of this call (USD)
Latency	Request latency (ms)
Throughput	Throughput, tokens generated per second (tokens per second)
Finish	Completion status (e.g., end_turn, tool_use, stop, max_tokens, length)
Action	Action button; click Details to view full request and response details

Viewing token details

Click the number in the Input Tokens column to open a popover showing a detailed token breakdown, including:

• prompt: Base input tokens
• input_cache_read: Tokens read from cache
• input_cache_write: Tokens written to cache
• input_cache_write_5_min: 5-minute cache write
• input_cache_write_1_h: 1-hour cache write

This helps you understand cache usage and optimize your caching strategy.

Viewing billing details

Hover over the number in the Cost column to open a popover showing billing details. What you see depends on your account type:

Pay-as-you-go users (Pay As You Go):

Shows a detailed breakdown by wallet category and settlement status:

• Purchased Credits: Funds topped up by the user; used when Reward Credits are insufficient
• Reward Credits: Credits such as top-up bonuses or insurance compensation; deducted first
• Status tags (e.g., Settled) indicate that the charge has been settled

Subscription users (Subscription): Shows the message: Included in your subscription usage - no separate charge.

Request Details page

Click Details on any log entry to view the full information for that call. The details page is split into two sections:

Left: Conversation content

The left panel displays the complete request and response content, split into Request and Response sections. Each section includes:

• User message: The input sent by the user
• System message: System prompt (if any)
• Assistant message: The response generated by the model
• Tool calls: If tool calling is used, the tool inputs and outputs are shown

Display mode toggle:

The left panel supports two display modes, switchable via the tabs at the top:

• Pretty mode (formatted): Presents the conversation in a formatted, readable view. Messages are grouped by role, and code is syntax-highlighted for easier reading and understanding.
• JSON mode (raw data): Shows the complete raw request and response data in JSON format, including all technical details—ideal for development debugging and troubleshooting.

Data source switching in JSON mode:

In JSON mode, you can further choose which stage of data to view:

• Request:
  • User -> ZenMux: The original request sent by the user to ZenMux
  • ZenMux -> Origin: The request ZenMux forwarded to the upstream provider
• Response:
  • ZenMux -> User: The response returned by ZenMux to the user
  • Origin -> ZenMux: The original response returned by the upstream provider to ZenMux

Switching display modes

Choose the display mode that best fits your needs:

• Use Pretty mode when reviewing the conversation and analyzing interaction quality
• Use JSON mode when debugging your API integration or troubleshooting technical issues, and switch data sources as needed to inspect request/response details at different stages

Right: Metrics and metadata

The right panel shows detailed technical metrics and metadata:

Model information (Model Information):

• Model: Model name used
• Provider: Model provider

Token counting (Token Counting):

• Input Tokens:
  • prompt: Prompt token count
  • input_cache_write: Tokens written to cache
  • input_cache_write_5_min: 5-minute cache write token count
  • input_cache_write_1_h: 1-hour cache write token count
  • input_cache_read: Tokens read from cache
• Output Tokens:
  • completion: Generated content token count
  • internal_reasoning: Internal reasoning token count
  • audio: Audio token count

Requests counting (Requests Counting):

• web_search: Number of web search requests

Multimodalities counting (Multimodalities Counting):

• image: Number of images processed

Performance metrics (Performance Metrics):

• First Token Latency (ms): Time from sending the request to receiving the first token (ms)
• Generation Time (ms): Time to generate the full response (ms)
• Throughput (tps): Token generation rate (tokens per second)

Raw metadata (Raw Metadata):

• Displays the full request metadata in JSON format
• Supports one-click copying of the JSON data

Raw Meta Information

In addition to the console UI, you can also use the raw Meta information returned by the API to understand cost details for a call. Below is a complete example:

Billing details JSON example

{
  "generationId": "2534CCEDTKJR00217635",
  "accountId": "2533AC0Q5MIe14613672",
  "model": "Anthropic: Claude Sonnet 4",
  "modelSlug": "anthropic/claude-sonnet-4",
  "apiKeyId": "2534AKkA4Nqn14642786",
  "providerSlug": "anthropic",
  "createdAt": "2025-08-22T02:49:18.000+00:00",
  "generationTime": 3298,
  "latency": 2177,
  "moderationLatency": 2169,
  "streamed": true,
  "cancelled": false,
  "throughput": "28.81",
  "nativeFeeTokens": {
    "input_cache_write_5_min": 0,
    "completion": 95,
    "input_cache_read": 0,
    "input_cache_write_1_h": 0,
    "prompt": 16527
  },
  "nativeTokens": {
    "completion_tokens": 95,
    "prompt_tokens": 16527,
    "total_tokens": 16622,
    "completion_tokens_details": { "reasoning_tokens": 0 },
    "prompt_tokens_details": {
      "ephemeral_1h_input_tokens": 0,
      "ephemeral_5m_input_tokens": 0,
      "cached_tokens": 0
    }
  },
  "finishReason": "tool_calls",
  "nativeFinishReason": "tool_calls",
  "originAmount": "0.051006",
  "billAmount": "0.051006",
  "discountAmount": "0",
  "realAmount": "0.051006",
  "nonpaymentAmount": "0",
  "ratingResponses": {
    "billAmount": 0.051006,
    "discountAmount": 0,
    "originAmount": 0.051006,
    "priceVersion": "0.0.3",
    "ratingDetails": [
      {
        "billAmount": 0,
        "discountAmount": 0,
        "feeItemCode": "input_cache_write_5_min",
        "originAmount": 0,
        "rate": 3.75
      },
      {
        "billAmount": 0.001425,
        "discountAmount": 0,
        "feeItemCode": "completion",
        "originAmount": 0.001425,
        "rate": 15
      },
      {
        "billAmount": 0,
        "discountAmount": 0,
        "feeItemCode": "input_cache_read",
        "originAmount": 0,
        "rate": 0.3
      },
      {
        "billAmount": 0,
        "discountAmount": 0,
        "feeItemCode": "input_cache_write_1_h",
        "originAmount": 0,
        "rate": 6
      },
      {
        "billAmount": 0.049581,
        "discountAmount": 0,
        "feeItemCode": "prompt",
        "originAmount": 0.049581,
        "rate": 3
      }
    ],
    "serviceCode": "anthropic#anthropic/ahtnropic_endpoint#anthropic/claude-sonnet-4"
  },
  "requestRetryTimes": 0,
  "finalRetry": true
}

Contact us

If you run into any issues while using ZenMux, or if you have suggestions or feedback, feel free to reach out via:

• Website: https://zenmux.ai
• Support email: [email protected]
• Business inquiries: [email protected]
• Twitter: @ZenMuxAI
• Discord community: http://discord.gg/vHZZzj84Bm

For more contact options and details, visit our Contact Us page.