Mastra adds time-to-first-chunk and time-to-first-token latency metrics
Overview
Mastra says users can now track time to first chunk (TTFC) and time to first token (TTFT) for every agent and model call in Mastra platform and Studio.
The metrics can be filtered and sorted in traces, spans, and threads, and queried through the Mastra CLI, Client SDK, and observability endpoints.
The metrics work on Mastra platform and on self-hosted projects whose storage.observability uses DuckDB, PostgreSQL, or ClickHouse. Requests need a platform access token and project ID. Mastra says the feature requires mastra@1.30.0 or later, and Studio requires mastra@1.37.0 and @mastra/core@1.81.0 or later.
Written by AI from the articles below · updated Nov 5, 12:13 PM ET
Check the sources:
Article timeline
The articles in this story. Times are ET.
- Mastra BlogOfficialMastra adds time-to-first-chunk and time-to-first-token latency metrics to observability
AIMastra says users can now track time to first chunk (TTFC) and time to first token (TTFT) for every agent and model call in Mastra platform and Studio. The metrics can be filtered and sorted in traces, spans, and threads, and queried through the Mastra CLI, Client SDK, and observability endpoints. They work on Mastra platform and self-hosted projects using DuckDB, PostgreSQL, or ClickHouse storage, and require mastra@1.30.0 or later.
Heat Trend
Not enough continuous observations to show a trend yet.