Product Positioning

telemetry.dev is an LLM observability platform for AI applications and agents, built on OpenTelemetry. It consolidates model calls, tool steps, and retrieval information from a single request or agent run into one trace, making it easy for developers to pinpoint failures, inspect latency, and track token usage and costs.

Confirmed Key Features

• Trace Tracking: View model calls, tool steps, and retrievals along with the duration of each step in a single trace.
• Cost and Performance Monitoring: Compare spend, latency, and errors by model, provider, and environment; costs are calculated at ingestion based on reported token usage and the model pricing catalog.
• MCP Analytics: View span counts, failure rates, and p95 latency of MCP traffic by method and tool.
• Alerts: Send email alerts for costs, budgets, agent loops, and model or tool errors.
• Prompt Management: Support versions, tags, and releases without redeployment; compare the performance of different versions in actual calls.
• Evaluation Tracing: Track each evaluation run along with its costs and errors.

Integration

First-party SDKs are available for TypeScript and Python, with integrations for providers and frameworks such as OpenAI, Anthropic, and Vercel AI SDK; projects with existing OpenTelemetry instrumentation can connect via OTLP/HTTP export. The platform also offers a read-only MCP server that allows AI agents to read workspace traces, logs, and usage.

Use Cases

Ideal for development teams that need to debug AI agent call chains, troubleshoot tool timeouts or model response issues, monitor token costs and latency, and manage prompt versions and evaluation results.

Pricing and Limits

The official website clearly offers a free plan: 30,000 ingest units per month (each span, log record, or metric data point counts as one unit), 14-day data retention, 1 project and 2 seats included, no credit card required. Pricing for other paid plans has not been confirmed in the collected materials.