The $26 plan is where the bill starts

Sentry is still the error tracker most teams install first. Its docs list 22 SDK platforms, from Python and Go to Unity, PlayStation and Xbox. When a release breaks, the issue page shows the exception, the line of code and the commit that shipped it. What has changed is the price. It is no longer one number.

The Team plan is $26/month and the Business plan $80/month, both "when billed annually with default pre-paid data" (checked on sentry.io/pricing on September 29, 2026). Those two figures buy a quota, not a product. Errors, spans, session replays, logs, profiling, cron monitors and uptime monitors are each metered separately. Every paid plan includes the same monthly allowance:

Two details in that table are easy to miss. Fifty replays is roughly one busy afternoon of checkout sessions. And Business does not raise any of these quotas: it buys SAML and SCIM, unlimited dashboards, advanced quota management and 13 months of sampled span retention. Its overage rates are about three times Team's - $0.0011125 per error against $0.0003625 in the 50k-100k band. Upgrading for SSO makes every extra event dearer.

The allowance has also shrunk. Sentry changed its plans on August 27, 2025, and according to a report by urgentry.com the included spans on Team and Business were cut from 10M to 5M at that point.

What a real team pays

Take a ten-developer team on Team with a moderately busy app: 150,000 errors, 25 million spans and 5,000 replays a month. Pay-as-you-go adds about $29 for errors, $40 for spans and $18.56 for replays. That is roughly $114/month, not $26. Turn on Seer, Sentry's AI agent, for all ten and add $400. The same volumes on Business come to about $265 before Seer, because the higher per-event rates eat the difference.

Review aggregators have collected worse stories. Last9 cites a Series B company paying $2,400 a month on a plan that started at $80.

Try Sentry free for 14 days

What Sentry does better than anyone

The core product earns its reputation. An unhandled exception arrives with its stack trace, the surrounding source lines, the release, the environment and the breadcrumbs that led to it. Sentry groups repeats into a single issue and alerts in Slack, Jira, Linear or email. Projects are unlimited on every plan, the free one included.

Tracing is where it pulls ahead of cheaper trackers. A slow request shows up as a waterfall from the browser click down to the database query. Newer SDKs also record AI agent spans - the model call, the tool it executed, how long each took. That is useful if you ship LLM features and want them in the same view as everything else.

Session Replay closes the loop on frontend bugs: you watch the user's session up to the moment the 500 arrived, with the error marked on the timeline. At 50 included replays a month, it is also the one most likely to run through your budget.

When the quota runs out, monitoring stops

This is the limit the pricing page does not put in bold. Sentry's own docs say that data sent after your reserved volume and pay-as-you-go budget are used up "will be dropped and you won't be charged for it. This means that you'll lose monitoring for the remainder of the billing cycle."

Think about when that happens. Quotas are exhausted by error storms, and error storms are incidents. The tool goes blind at the moment you need it most, and stays blind until the month turns over.

The pay-as-you-go budget does not help as much as it sounds. It is "shared among all categories on a first-come, first-served basis", so a spike in replays can spend the money your errors needed. Spike protection, enabled per project by an owner or billing member, stops a flood from running up a bill, but it does so the same way: by dropping events. Unused reserved volume expires at the end of each month, and downgrades and cancellations take effect at the end of the contract "and cannot be refunded". Buy annual reserved volume cautiously.

Retention also varies by meter. Errors and replays keep 90 days on Team and Business, but logs, profiles and crons keep 30 days on every plan, Business included. If you want Sentry to replace a log store, that is a short window.

Seer: the AI agent billed per contributor

Seer reads an issue's stack trace, traces, logs and profiles, proposes a root cause and can open a pull request with a fix. It can also hand the fix to an outside coding agent. It answers natural-language questions about your telemetry and reviews pull requests on GitHub and GitLab.

Since January 2026 it costs $40 per active contributor per month on Team or Business, billed on top of the plan and outside the pay-as-you-go budget. An active contributor is anyone who opens two or more pull requests to a Seer-enabled repository. The previous pricing - $20 a month per subscription plus $25 of credits, with fixes at $1 a run - is gone. On GitHub, bot accounts are excluded from the count; on GitLab, Sentry says every group member counts, bots included.

There are three more limits. Seer works only with GitHub.com and GitLab.com, not GitHub Enterprise Server or self-managed GitLab. It is not available on the free plan. And it does not exist in self-hosted Sentry at all. Sentry states that Seer "does not use your application data, error information, or source code to train AI models."

For a team that triages a lot of production errors, $40 a head can pay for itself in saved hours. For a team of fifteen where only three people touch incidents, the per-contributor count is the wrong unit, and the $600 a month is hard to defend.

The free plan and self-hosting

The free Developer plan gives you 5,000 errors, 5 GB of logs, error monitoring and tracing, and 30 days of history. Its real limit is one line on the pricing page: "Limited to one user." It suits a side project. It does not suit a two-person startup, which lands on Team the day the second developer joins. New accounts do get a 14-day trial with Team retention.

Self-hosting is free in licence terms and expensive in machine terms. The published minimum is 4 CPU cores, 16 GB of RAM plus 16 GB of swap and 20 GB of disk, and Sentry recommends 32 GB of RAM. The stack runs Postgres, Redis, Kafka and ClickHouse under Docker Compose, and Sentry says it offers no guidance on scaling it up. The code is under the Functional Source License, which turns into Apache 2.0 two years after each release and forbids reselling Sentry as a service. You lose spike protection, the choice of data region and Seer. Sentry describes self-hosted as "the Business plan without any software limitations and no paid tier" - true, provided someone on your team enjoys running Kafka.

On the hosted side, data lives in the US (Iowa) or the EU (Frankfurt), and Sentry holds SOC 2 and ISO 27001 reports. A HIPAA Business Associate Amendment exists; confirm with sales which plan it comes with.

Cheaper alternatives

GlitchTip accepts the Sentry SDK protocol, so moving is mostly a matter of changing the DSN in your config. It is free up to 1,000 events with unlimited members, $15/month for 100k events and $50/month for 500k. You give up tracing depth, Session Replay and Seer. For a team that mostly needs "tell me when the backend throws", the saving is real: the 150,000 errors in our example fit GlitchTip's $50 tier.

PostHog gives away the first 100,000 exceptions a month - twice Sentry's paid allowance - and then charges $0.00037 each, close to Sentry's Team rate. It makes sense if you already use PostHog for product analytics. It has no uptime monitoring yet.

If you need metrics, logs and traces in one place more than you need issue grouping, read our Grafana Cloud review.

Verdict

Buy Sentry if you ship code across several platforms, especially mobile or games, and want the best stack-trace-to-commit workflow available. Stay on Team, set a pay-as-you-go budget, turn on spike protection for every project and treat Session Replay as a sampled feature, not a default. Add Seer only for the people who actually fix production issues.

Skip it if you are a small team that needs error alerts and little else: GlitchTip does that for $15 a month with the same SDKs. Skip it also if your error volume swings wildly and you cannot tolerate monitoring going dark mid-month. There, PostHog's larger free allowance or a self-hosted GlitchTip gives you more headroom per dollar.

Sentry is still the best error monitor on the market. It is no longer the cheap one, and the $26 on the pricing page is the least you will pay, not what you will pay.

Visit Sentry