# pydantic/logfire

AI observability platform for production LLM and agent systems.

Repository: https://github.com/pydantic/logfire
Canonical: https://ross.abutalabs.com/products/logfire
Homepage: https://pydantic.dev/logfire/
Language: Python
License: MIT
License Family: permissive
Topics: fastapi, logging, observability, openai, opentelemetry, pydantic, python, trace, metrics, ai, ai-tools, pydantic-ai, agent-observability, ai-observability, evals, llm-observability
Last push: 2026-08-26T22:39:14+00:00

## Health v2 (maintenance only)
Score: 87/100 (v2, computed 2026-09-03T02:39:23.370411+00:00)
- activity 99, release rhythm 86, longevity 61
- inputs: {"age_days": 862, "days_push": 7, "days_rel": 13, "gap_med": 3.0, "n_releases_24m": 135}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 4441, forks 282 (observed 2026-08-28T04:08:49.376155+00:00)

## What it is
Pydantic Logfire is an AI observability platform built on OpenTelemetry that traces LLM calls, agent runs, and application telemetry in one correlated timeline, queryable with SQL. It includes evaluation tooling (datasets, experiments, live evals, human review) and a Python SDK with integrations for Pydantic AI, OpenAI, Anthropic, FastAPI, and other OTel-instrumented stacks.

## Use cases
- trace llm calls and agent runs in production
- monitor ai application quality with evals
- debug slow agent tool calls and model requests
- observe python app traces logs and metrics with opentelemetry
- run offline and online evaluations of llm systems
- review and annotate agent runs by hand
- query telemetry data with sql
- instrument openai or anthropic sdk calls

## When to choose
- you want unified observability for both your backend and LLM/agent stack in one trace
- you use Python, Pydantic, Pydantic AI, or FastAPI and want first-class instrumentation
- you need production evals, live monitoring, and human review of agent outputs
- you prefer OpenTelemetry compatibility and SQL-based querying of telemetry

## When to avoid
- you need a fully self-hosted open-source observability stack with no hosted component (the server is proprietary; only the SDK is MIT)
- your stack is non-Python and you want deep language-specific insights
- you need a free unlimited-volume solution for very high telemetry volumes beyond 10M records/month
- you want a vendor-neutral tool you can point at any existing OTel backend without a Logfire account

## Facets
- artifact type: service
- maturity: active
- function: monitoring, tracing, logging, alerting, analytics, llm-inference, agent-framework, rag, sdk, mcp
- domain: monitoring, large-language-models, developer-tools, data-visualization
- platform: python, rust, cloud, self-hosted, cross-platform
- tags: opentelemetry, llm-observability, agent-observability, evals, pydantic-ai, fastapi, openai, anthropic, sql-querying, hosted-platform, freemium, ai-agents, retrieval-augmented-generation, nodejs, web-server, docker

## Member repositories
- pydantic/logfire (main) score 87

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:08:49.376155+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T18:20:50.201110+00:00, confidence not recorded.
  - readme: https://github.com/pydantic/logfire (fetched 2026-08-28T04:08:49.376155+00:00, sha 1983205fe553)
  - homepage: https://pydantic.dev/logfire/ (fetched 2026-08-29T09:08:07.778340+00:00, sha 146d1f7c7a27)
  - site_page: https://pydantic.dev/docs/logfire/evaluate/evals (fetched 2026-08-29T09:08:07.790348+00:00, sha 2d2ab3260f82)
  - site_page: https://pydantic.dev/docs/logfire/evaluate/datasets-and-experiments (fetched 2026-08-29T09:08:07.792147+00:00, sha 2d2ab3260f82)
  - site_page: https://pydantic.dev/docs/logfire/evaluate/human-review (fetched 2026-08-29T09:08:07.793816+00:00, sha c1d3fbb0535c)
  - site_page: https://pydantic.dev/docs/logfire/evaluate/live-evals (fetched 2026-08-29T09:08:07.795568+00:00, sha 1af42e47ff52)
  - site_page: https://pydantic.dev/docs/logfire/integrations/llms/pydanticai (fetched 2026-08-29T09:08:07.797334+00:00, sha 8b503d3197ec)
  - site_page: https://pydantic.dev/docs/logfire/integrations/llms/openai (fetched 2026-08-29T09:08:07.798968+00:00, sha b86a70412188)
  - site_page: https://pydantic.dev/docs/logfire/integrations/llms/anthropic (fetched 2026-08-29T09:08:07.800578+00:00, sha 2bb264194064)
  - registry_pypi: https://pypi.org/pypi/logfire/json (fetched 2026-08-29T09:08:07.802481+00:00, sha 33099b5f17c1)
- Data as of 2026-08-30T08:39:29.467469+00:00.