telemetry.dev
LLM observability platform that traces model calls, tool steps, and retrievals to inspect token usage, cost, latency, and errors
An LLM observability platform that traces model calls, tool steps, and retrievals to debug failures and monitor token usage, cost, latency, and errors in AI applications. It sells to developers, data teams, and operations teams building AI agents and applications at B2B companies. It is delivered as SaaS and infrastructure API via TypeScript and Python SDKs or any existing OpenTelemetry OTLP/HTTP exporter, normalizing traces from any framework or language into a consistent view.
Key features
- Trace model calls, tool steps, retrievals
- Inspect token usage and cost
- Monitor latency and p95
- Track error rate and status
- Filter traces by status and environment
- Group cost and latency by model
- Group by provider and environment
- Capture inputs and outputs per environment
- Redact input and output fields
- Calculate cost per call from tokens
- Track token trends over time
- Ingest via OTLP/HTTP endpoint
- TypeScript and Python SDKs
- Integrations for OpenAI and Anthropic
- Integration for Vercel AI SDK
- Integration for LangChain
Social posts
- No social media activity detected
GTM channels
- API
- Docs
ICP
- Software developers
- Engineering teams
- DevOps sre teams