text stringlengths 0 59.1k |
|---|
- 1 day, 2 days, 3 days, 7 days, 14 days, 30 days |
- Custom date range |
Free plan users can access up to 7 days of data. 30-day analytics requires a Pro plan. |
## Overview Section |
### Trace Metrics Card |
Displays aggregate statistics for all traces in the selected period: |
| Metric | Description | |
| ------------ | -------------------------------------------------- | |
| Total Traces | Number of agent/workflow executions | |
| Success Rate | Percentage of traces that completed without errors | |
| Error Count | Number of failed traces | |
| Avg Duration | Average execution time across all traces | |
### LLM Call Metrics Card |
Summarizes LLM API call statistics: |
| Metric | Description | |
| ------------ | ---------------------------------------- | |
| Total Calls | Number of LLM API requests | |
| Success Rate | Percentage of successful LLM calls | |
| Failed Calls | Number of LLM calls that returned errors | |
| Avg Latency | Average response time for LLM calls | |
### Trace Counts Chart |
A time-series chart showing trace volume over time. Data points are color-coded by status: |
- **Green**: Successful traces |
- **Red**: Failed traces |
- **Yellow**: In-progress traces |
Click on a data point to filter traces by that time period. |
### LLM Call Rates Chart |
Shows LLM call volume over time, broken down by: |
- Success vs failure |
- Model provider (OpenAI, Anthropic, etc.) |
## Cost & Usage Section |
### LLM Cost Metrics Card |
Displays cost breakdown for LLM usage: |
| Metric | Description | |
| ------------- | ---------------------------------------- | |
| Total Cost | Sum of all LLM API costs | |
| Cost by Model | Breakdown by model (GPT-4, Claude, etc.) | |
| Cost Trend | Comparison with previous period | |
### User Cost Metrics Card |
Shows cost attribution by user: |
| Metric | Description | |
| ------------- | ---------------------------- | |
| Top Users | Users with highest LLM costs | |
| Cost per User | Average cost per active user | |
| User Count | Number of unique users | |
### Trace Cost Chart |
Time-series chart of costs per trace execution. Helps identify: |
- Cost spikes |
- Expensive workflows |
- Cost trends over time |
### Token Usage Chart |
Displays token consumption over time: |
- **Prompt tokens**: Input tokens sent to LLM |
- **Completion tokens**: Output tokens received from LLM |
- **Total tokens**: Combined usage |
## Latency Section |
### Trace Latency Chart |
Shows end-to-end execution time for traces: |
- P50 (median) |
- P90 (90th percentile) |
- P99 (99th percentile) |
- Average |
### LLM Latency Chart |
Displays LLM response times: |
- By model |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.