Real-time performance & usage tracking
Real-Time Telemetry for Real-Time Agents
Monitor, analyze, and optimize your AI workflows in real-time—no guesswork, just results.
Get Started
Requests over time (Success Vs Failure)
Requests over time (Success Vs Failure)
1k
800
600
400
200
00
Time
Success
Failure
Request Count
Cost Over Time
Cost Over Time
1k
400
200
00
Time
Input cost
Output cost
total cost
Cost (USD)
Model Leaderboard
Model .svg) | Requests | Total Tokens | Avg Latency | Total Cost | Success rate |
|---|---|---|---|---|---|
gr-test | 1 | 0 | 0 ms | $0.0001 | 0.00% |
Claude | 36 | 0 | 0 ms | $0.0001 | 0.00% |
groq - 1 | 155 | 65935 | 174 ms | $0.0001 | 0.00% |
groq - 2 | 123 | 60734 | 256 ms | $0.0001 | 0.00% |
g-gen-ai | 30 | 1395 | 412 ms | $0.0002 | 0.00% |
google-generative-ai | 300 | 99002 | 474 ms | $0.0095 | 0.00% |
google-gen-ai | 450 | 148481 | 490 ms | $0.0143 | 0.00% |
Open-ai | 31 | 1963 | 597 ms | $0.0016 | 0.00% |
amz-bedrock | 123 | 44994 | 704 ms | $0.0000 | 0.00% |
groq - 3 | 477 | 247170 | 1494 ms | $0.0136 | 0.00% |
Tokens Over time
Tokens Over time
1k
600
400
200
00
Time
Input Tokens
Output Tokens
Token Count
.avif)
.avif)
Precision Performance Insights
Engineering-Grade Telemetry, Natively Integrated
Native Agent Telemetry with Fine-Grained Tracing
FloTorch automatically instruments every agent and tool execution with detailed telemetry—track inputs, outputs, latencies, retries, and errors at each step of your workflow.
.png)
Live Debug Panels with Step-Level Replay
Reproduce failures instantly with FloTorch’s visual replay of live sessions—drill into each tool call, API response, and intermediate state without re-running agents.
.png)
Integrated with FloTorch’s Control Plane
Performance insights are fully integrated with FloTorch's orchestration engine—link performance regressions to recent deployment changes or config diffs instantly.
Real-Time Token, Cost, and Latency Metrics by Agent & Chain
Understand exactly how each agent and chain performs: token consumption, cost per run, latency breakdowns, and throughput—all streamed in real time via FloTorch’s observability layer.
Workflow-Specific Performance Dashboards
FloTorch auto-generates dashboards per workflow or environment, exposing health, usage trends, and error rates across production, staging, or test deployments
.png)
Extensible with OpenTelemetry Export
Export telemetry data to your own Grafana or Prometheus setup, or feed into custom monitoring pipelines via FloTorch’s native OpenTelemetry support.
