Request Logs

Request logs provide detailed visibility into every API request made through LangMart. Use them to debug issues, analyze usage patterns, and understand costs.

Viewing Request Logs

Dashboard View

Navigate to Analytics > Request Logs in the sidebar
View the list of recent requests with summary information
Click on any request to see full details

API Access

# List request logs
curl -X GET "https://api.langmart.ai/api/account/request-logs?limit=50" \
  -H "Authorization: Bearer YOUR_API_KEY"

# Get enhanced logs with token usage
curl -X GET "https://api.langmart.ai/api/account/request-logs/enhanced" \
  -H "Authorization: Bearer YOUR_API_KEY"

# Get specific log details
curl -X GET "https://api.langmart.ai/api/account/request-logs/{log_id}/details" \
  -H "Authorization: Bearer YOUR_API_KEY"

Log Details

Basic Information

Each log entry contains:

Field	Description
`id`	Unique request identifier
`created_at`	Timestamp of the request
`method`	HTTP method (POST, GET)
`endpoint`	API endpoint called
`response_status`	HTTP status code
`ip_address`	Client IP address
`user_agent`	Client user agent

Token Usage

Enhanced logs include detailed token metrics:

Field	Description
`input_tokens`	Number of tokens in the request
`output_tokens`	Number of tokens in the response
`total_tokens`	Total token count
`total_cost`	Total cost in USD
`platform_fee`	LangMart platform fee
`provider_revenue`	Amount paid to provider

Latency Metrics

Field	Description
`latency_ms`	Total request latency in milliseconds
`time_to_first_token_ms`	Time until first token (streaming)

Model Information

Field	Description
`model_name`	Model identifier (e.g., `gpt-4o`)
`model_display_name`	Human-readable model name
`provider_name`	Provider name (e.g., OpenAI)
`provider_key`	Provider identifier

Enhanced Logs with Resolution Journey

When requests involve fallbacks or retries, the resolution_attempts field provides a complete picture:

{
  "resolution_attempts": {
    "summary": {
      "totalCandidatesTried": 2,
      "successfulCandidate": "anthropic/claude-3-sonnet",
      "totalDuration": 1250
    },
    "attempts": [
      {
        "candidate": "openai/gpt-4o",
        "status": "failed",
        "error": "rate_limit_exceeded",
        "duration": 450
      },
      {
        "candidate": "anthropic/claude-3-sonnet",
        "status": "success",
        "duration": 800
      }
    ]
  }
}

Resolution Journey Fields

Field	Description
`totalCandidatesTried`	Number of models attempted
`successfulCandidate`	Model that handled the request
`totalDuration`	Total time including retries
`attempts`	Array of individual attempts

Filtering and Searching

Available Filters

# Filter by response status
curl "https://api.langmart.ai/api/account/request-logs?status=200"

# Filter by model
curl "https://api.langmart.ai/api/account/request-logs/enhanced?model=gpt-4o"

# Filter by provider
curl "https://api.langmart.ai/api/account/request-logs/enhanced?provider=openai"

# Filter by date range
curl "https://api.langmart.ai/api/account/request-logs?start_date=2025-01-01&end_date=2025-01-31"

# Filter by HTTP method
curl "https://api.langmart.ai/api/account/request-logs?method=POST"

# Filter by endpoint
curl "https://api.langmart.ai/api/account/request-logs?endpoint=/v1/chat/completions"

# Filter by cost range
curl "https://api.langmart.ai/api/account/request-logs/enhanced?min_cost=0.01&max_cost=1.00"

# Filter by latency
curl "https://api.langmart.ai/api/account/request-logs/enhanced?min_latency=1000"

# Filter for errors only
curl "https://api.langmart.ai/api/account/request-logs/enhanced?has_error=true"

# Filter for requests with retries
curl "https://api.langmart.ai/api/account/request-logs/enhanced?has_retries=true"

Pagination

# Get first 50 logs
curl "https://api.langmart.ai/api/account/request-logs?limit=50&offset=0"

# Get next 50 logs
curl "https://api.langmart.ai/api/account/request-logs?limit=50&offset=50"

Response includes pagination metadata:

{
  "success": true,
  "data": [...],
  "pagination": {
    "total": 1250,
    "limit": 50,
    "offset": 0,
    "returned": 50,
    "has_more": true,
    "next_offset": 50
  }
}

Log Statistics

Get aggregated statistics about your requests:

curl -X GET "https://api.langmart.ai/api/account/request-logs/stats?start_date=2025-01-01" \
  -H "Authorization: Bearer YOUR_API_KEY"

Response includes:

{
  "success": true,
  "data": {
    "period": {
      "start_date": "2025-01-01T00:00:00Z",
      "end_date": "2025-01-31T23:59:59Z"
    },
    "summary": {
      "total_requests": 15420,
      "days_with_requests": 31,
      "successful_requests": 15180,
      "client_errors": 180,
      "server_errors": 60,
      "unique_models_used": 8,
      "unique_endpoints": 3,
      "success_rate": "98.44%"
    },
    "top_models": [
      {"model_name": "gpt-4o", "request_count": 8500},
      {"model_name": "claude-3-sonnet", "request_count": 4200}
    ],
    "top_endpoints": [
      {"endpoint": "/v1/chat/completions", "request_count": 14800}
    ],
    "daily_counts": [
      {"date": "2025-01-31", "request_count": 520, "successful_count": 512, "error_count": 8}
    ]
  }
}

Viewing Full Request/Response

For debugging, you can view the complete request and response:

curl -X GET "https://api.langmart.ai/api/account/request-logs/{log_id}/details" \
  -H "Authorization: Bearer YOUR_API_KEY"

Note: Sensitive headers (Authorization, API keys) are automatically redacted in the response.

Example Response

{
  "success": true,
  "data": {
    "id": "abc123",
    "request_body": {
      "model": "gpt-4o",
      "messages": [{"role": "user", "content": "Hello"}]
    },
    "response_body": {
      "choices": [{"message": {"content": "Hi there!"}}],
      "usage": {"prompt_tokens": 10, "completion_tokens": 5}
    },
    "request_headers": {
      "content-type": "application/json",
      "authorization": "***REDACTED***"
    },
    "input_tokens": 10,
    "output_tokens": 5,
    "total_cost": 0.0001,
    "latency_ms": 450
  }
}

Connection Pool Information

For requests using connection pools, logs include pool details:

Field	Description
`connection_pool_id`	Pool identifier
`connection_pool_name`	Pool display name
`used_connection_id`	Specific connection used
`used_connection_name`	Connection display name
`pool_strategy`	Load balancing strategy

Best Practices

Debugging Failed Requests

Filter for error responses: has_error=true
Check the error_code, error_type, and error_message fields
Review the request body for invalid parameters
Check resolution attempts if fallbacks were configured

Monitoring Costs

Use the enhanced logs endpoint for cost data
Filter by date range for billing periods
Group by model to identify expensive operations
Set up cost threshold alerts

Performance Analysis

Filter by min_latency to find slow requests
Check time_to_first_token_ms for streaming performance
Identify patterns in latency by model or time of day
Review resolution journey for retry impact

Usage Analytics - Aggregated usage trends
Error Tracking - Error analysis and debugging
API Errors - Error code reference

Previous Error Tracking Next Usage Analytics