Pydantic Logfire
Pydantic Logfire is an OpenTelemetry-native platform with a closed-source server and MIT client SDKs, priced per million telemetry records, offering EU or US regions and a 10M-record free tier.
Last verified
| Licence | proprietary (SaaS core) with MIT-licensed client SDKs — vendor states the server application and UI are closed source; the pydantic/logfire SDK repo (the same product's own client libraries) carries the MIT License Source, checked pydantic.devhttps://pydantic.dev/logfireChecked |
|---|---|
| Self-host | partial: no on-prem option on Personal, Team or Growth (the plan comparison shows '—' in the On-prem row for all three); self-hosting is an Enterprise-plan deployment, run from an open-sourced Helm chart on the customer's own Kubernetes cluster with Postgres and an S3-compatible backend Source, checked pydantic.devhttps://pydantic.dev/pricingChecked |
| OpenTelemetry | yes — OpenTelemetry-native: ingests OTLP/OpenTelemetry signals and any OTel-instrumented framework works without a Logfire-specific integration Source, checked pydantic.devhttps://pydantic.dev/logfireChecked |
| Free tier limits | Personal plan, free forever: 10 million telemetry records/month (logs, spans and metrics) hard-capped at $0, 1 seat + 2 read-only guests, 3 projects, 30-day retention Source, checked pydantic.devhttps://pydantic.dev/pricingChecked |
| Price per 100k traces | not trace-denominated: priced per million records (logs, spans and metrics). On Team and Growth, usage above the $20/mo credit (10M records) is $2 per million, i.e. $2/1,000,000 x 100,000 = $0.20 USD per 100k records; Personal allows no additional usage (the comparison table's Additional usage row reads 'Not allowed' and ingestion pauses at 10M records), and Enterprise Cloud/Dedicated/Self-hosted additional usage is 'Custom'. A trace contains many records, so no per-trace figure is published. Source, checked pydantic.devhttps://pydantic.dev/pricingChecked |
| Data residency | EU or US regions on Personal, Team, Growth and Enterprise Cloud; Enterprise Dedicated in any GCP region; Enterprise Self-hosted on premise (the On-prem row is '—' for Personal, Team and Growth) Source, checked pydantic.devhttps://pydantic.dev/pricingChecked |
| Retention | Personal 30 days; Team 30 days; Growth up to 90 days; Enterprise custom (Cloud 'Longer than 90 days', Dedicated custom, Self-hosted customer determined / multi-year) Source, checked pydantic.devhttps://pydantic.dev/pricingChecked |
| SDK languages | First-party SDKs: Python, JavaScript/TypeScript (Node.js, browsers, Cloudflare Workers, Deno) and Rust; all other languages (Go, Java, .NET, Ruby, PHP, Elixir) via standard OpenTelemetry SDKs Source, checked pydantic.devhttps://pydantic.dev/logfireChecked |
| Last release | v4.41.0, released 2026-08-20 (Logfire release notes; the same entry heads the pydantic/logfire CHANGELOG) Source, checked raw.githubusercontent.comhttps://raw.githubusercontent.com/pydantic/logfire/main/CHANGELOG.mdChecked |
Assessment
The vendor states the split plainly: the SDKs are open source, the server application and UI closed source, and the client SDK repository carries the MIT licence. Self-hosting is an Enterprise deployment from an open-sourced Helm chart on the customer's Kubernetes cluster with Postgres and an S3-compatible backend; no on-premises option exists below Enterprise.
OpenTelemetry is the substrate: Logfire ingests OTLP signals, and any OTel-instrumented framework works without a Logfire-specific integration. First-party SDKs are Python, JavaScript/TypeScript (Node.js, browsers, Cloudflare Workers, Deno) and Rust; everything else arrives via standard OTel SDKs.
The unit is a telemetry record, counting logs, spans and metrics alike, so a trace consumes many and no per-trace price is published. The free Personal plan carries 10 million records a month hard-capped at $0, one seat, two read-only guests and three projects; the vendor says it will "eventually stop ingesting new data" if usage goes "well over the free allowance". On Team and Growth, usage beyond the $20 monthly credit is $2 per million records, or $0.20 per 100k; Enterprise is custom.
Data residency is EU or US on Personal, Team, Growth and Enterprise Cloud, any GCP region on Enterprise Dedicated, and on premises when self-hosted. Retention is 30 days on Personal and Team, up to 90 on Growth. The latest release is v4.41.0 (20 August 2026).
Something wrong on this page? Suggest a correction — corrections are free and open to everyone.