Whether you are weighing a Sentry alternative or choosing your first error tracking stack: Logfire brings errors, agent runs, and application telemetry into one OpenTelemetry-native workflow with PostgreSQL-compatible SQL, pending spans, and a path from production evidence to evaluation and improvement.
Trusted by teams building production software and AI
| Feature | Logfire | Sentry |
|---|---|---|
| Primary Focus | OpenTelemetry-native full-stack and AI observability | Error, performance, and agent monitoring |
| AI/LLM Support | Automatic framework instrumentation and agent traces | Dedicated Agent Monitoring integrations and AI trace views |
| Agent context | Agent spans inside the surrounding application trace | Agent runs, AI spans, and conversations |
| Logging | Structured logs queryable with traces and metrics | Structured logs correlated with traces and issues |
| In-flight work | Live View for pending spans | Recorded agent traces and conversation views |
| Query Interface | Full PostgreSQL-compatible SQL over traces, logs, and metrics: joins, subqueries, window functions, plus MCP for agents | Field filters (span.op:http.client) with a fixed aggregate menu (p50-p95, avg, max) in Trace Explorer |
| Cross-signal context | AI, browser, database, API, service, log, metric, and infrastructure telemetry | Frontend, backend, agent spans, errors, and logs |
| Pricing | Personal: 10M records free; Team: $49/month + $2/M over 10M | Developer free; Team $26 + reserved volume or a configured pay-as-you-go budget; data drops at the limit |
Logfire keeps model calls, tools, browser and backend spans, databases, APIs, logs, metrics, and infrastructure in one investigation. Query the telemetry with PostgreSQL-compatible SQL and MCP, then jump from a browser trace to the exact moment in its session replay.
Logfire carries production traces into hosted LLM judges, deterministic evaluators, annotation queues, scheduled optimization proposals, and model-swap recommendations. Teams can investigate a run, review the evidence, and apply the next change from the same workflow.
Logfire's pending spans show active model calls and tool executions while an agent run is still in progress. That lets a developer inspect a slow or stuck operation before the request completes, alongside the completed-run analysis available after it finishes.
Sentry's Trace Explorer queries spans with field:value filters and a fixed menu of aggregates (p50-p95, avg, max). Logfire gives you the whole SQL language over traces, logs, and metrics: "Show me all FastAPI requests that called our LLM more than 3 times." "Which deploy introduced the slow queries, and what did those requests log?" Joins, subqueries, and window functions are all available, and Logfire's MCP server gives coding agents the same interface, so they can investigate production behavior without a dashboard being predefined for each question.
Start free with 10 million spans, logs, and metrics each month. No credit card required.
| Illustrative monthly workload | Logfire | Sentry |
|---|---|---|
| Base-plan prices | $49 Team; 10M records; 30 days | $26 Team; independent quotas; up to 90 days |
| 1M errors/month | $49/month; errors are not a separate meter and 1M error records sit inside Team's included 10M | $225.40/month |
| 10M errors/month | $49/month; 10M error records equal Team's included allowance | $1,912.90/month |
Calculation scope
Three reproducible Sentry Team calculations using the public calculator's included quotas and pay-as-you-go rates, with the Logfire total computed for the same workload on the equivalent paid plan. The base-plan row is not presented as an equivalent configuration. The error examples need no cross-meter conversion because Logfire has no separate error meter.
Basis: USD; US annual Team list price. Prices checked .
The official calculator starts with 50,000 errors, 5 GB logs, and 5 million spans, with no pay-as-you-go usage.
$26 + $0 additional usage $26 The first 50,000 errors are included, then each published error tier is applied only to the volume inside that band. Logfire has no error meter, so 1M error records sit inside Team's included 10M and add nothing to the base. Both sides are quoted on their paid team plan; the free Personal plan would also cover this volume, but it pauses ingestion at its limit and leaves no allowance for the traces, logs and metrics this page argues errors belong beside.
$49 + (1M error records within Team's included 10M) × $0 $49 $26 $26 50,000 × $0.0003625 $18.13 400,000 × $0.0002188 $87.52 500,000 × $0.0001875 $93.75 Sentry's published tiers applied to the full band widths. In Logfire, 10M error records exactly consume Team's included allowance, with no error-specific rate at any volume. This is the ceiling of the row rather than headroom: any traces, logs or metrics alongside these errors bill at the published per-record overage.
$49 + (10M error records = Team's included 10M) × $0 $49 $26 $26 50,000 × $0.0003625 $18.13 400,000 × $0.0002188 $87.52 9,500,000 × $0.0001875 $1,781.25 “Covers 10 million logs, spans, and metrics every month.”
The official calculator shows 20 million records and no extra seats totaling $69/month on Team.
“Any usage above your quota is billed at pay-as-you-go rates.”
The default annual Team calculator shows $26/month with 50,000 errors, 5 GB logs, and 5 million spans included.
“Any data sent after you've run through your reserved volume and PAYG budget will be dropped.”
Sentry documents prepaid reserved volume, an optional pay-as-you-go budget, and dropping after both are exhausted.
“When spike protection is triggered, events start getting dropped.”
Sentry documents independent quotas and spike protection that can drop excess events.
If you are looking for a Sentry alternative that combines error tracking and error monitoring with full traces, logs, metrics, and agent observability, Logfire covers that in one OpenTelemetry-native product with usage-based pricing and a free tier.
For teams that want errors, full-stack telemetry, and agent observability in one product, yes. Logfire covers errors, logs, traces, metrics, browser RUM and session replay, and agent monitoring in one OpenTelemetry-native product. If Sentry is already embedded in your workflow, you can run both while you move instrumentation and investigations to Logfire.
Yes. Error tracking is built in: Logfire captures exceptions as part of the same trace as model calls, tool calls, database queries, and API requests, with correlated logs and metrics available to the same investigation.
Yes. Logfire's Live View shows pending spans before a request completes, so you can inspect where an active agent is spending time instead of waiting for the request to finish.
Logfire connects agent traces to PostgreSQL-compatible SQL, MCP investigation, pending-span Live View, hosted and code-based evaluators, annotation queues, scheduled optimization proposals, model-swap recommendations, and managed agent configuration. The same investigation includes the application and infrastructure around the agent.
Sentry's Trace Explorer filters spans with field:value syntax such as span.op:http.client and offers a fixed menu of aggregates (p50-p95, avg, max), grouped by span attributes. Logfire exposes traces, logs, and metrics as PostgreSQL-compatible tables, so an investigation can use joins, subqueries, window functions, and arbitrary SQL expressions - for example correlating slow spans with a deploy version and the error logs from the same requests in one query. Coding agents get the same interface through Logfire's MCP server.
Sentry documents that events sent after your reserved volume and any pay-as-you-go budget are exhausted are dropped, and spike protection can drop excess events sooner. Logfire's paid plans bill records beyond the included 10M at a single published overage rate instead of discarding data at a quota boundary.
Yes. You can keep an existing Sentry deployment while introducing Logfire for OpenTelemetry-native full-stack and agent observability, then move workflows incrementally.
Get started with 10 million free spans, logs, and metrics per month. No credit card required.