Sentry's pricing page leads with $26 a month, and for a small team that number is real. It covers 50,000 errors, 5 million spans and 5 GB of logs. What the headline hides is how the bill scales once you pass those quotas, and how differently the Team and Business plans charge for the same error.
Short answer: Most teams should buy Team at $26 a month on annual billing, set a pay-as-you-go budget with a hard cap, and turn on spike protection before anything else. Move to Business at $80 only when you need SAML and SCIM. Business charges about three times as much per extra error, so a noisy app gets expensive fast. Self-hosting is free in license terms, but it needs a 16 GB server and someone who enjoys running Kafka.
Prices below come from Sentry's pricing page and its pricing documentation, checked on October 31, 2026.
The plans
| Plan | Annual billing | Monthly billing | Users | Errors included | Data lookback |
|---|---|---|---|---|---|
| Developer | Free | Free | 1 | 5,000 | 30 days |
| Team | $26/mo ($312/yr) | $29/mo | Unlimited | 50,000 | Up to 90 days |
| Business | $80/mo ($960/yr) | $89/mo | Unlimited | 50,000 | Up to 90 days |
| Enterprise | Custom | Custom | Unlimited | Custom | Custom |
Team and Business include the same base volumes: 50,000 errors, 5 million spans, 5 GB of logs, 5 GB of application metrics, 50 session replays, one cron monitor, one uptime monitor and 1 GB of attachments.
What Business adds is mostly administration: SAML and SCIM, unlimited custom dashboards (Team gets 20), and advanced quota management. Team adds unlimited users, API access and third-party integrations over the free plan.
The free Developer plan is genuinely useful for a side project, but it's limited to one user. The moment a second engineer needs access, you're on Team.
How overages work
Every paid plan lets you buy extra volume two ways: reserved (prepaid, cheaper) or pay-as-you-go (PAYG, billed after use). Error pricing is tiered, and the gap between Team and Business is the most important number in this post:
| Errors per month | Team reserved | Team PAYG | Business reserved | Business PAYG |
|---|---|---|---|---|
| 50k to 100k | $0.00029 | $0.00036 | $0.00089 | $0.00111 |
| 100k to 500k | $0.000175 | $0.00022 | $0.0005 | $0.000625 |
| 500k to 10M | $0.00015 | $0.00019 | $0.0003 | $0.000375 |
| 10M to 20M | $0.00013 | $0.00016 | $0.00026 | $0.000325 |
| Over 20M | $0.00012 | $0.00015 | $0.00024 | $0.0003 |
Two worked examples, error volume only:
200,000 errors a month
- ▸Team: $26 + $32 reserved = $58/month (or $66 on PAYG)
- ▸Business: $80 + $94.50 reserved = $174.50/month (or about $198 on PAYG)
2 million errors a month
- ▸Team: $26 + $309.50 reserved = about $336/month (about $413 on PAYG)
- ▸Business: $80 + $694.50 reserved = about $775/month (about $948 on PAYG)
At 2 million errors, Business costs more than twice as much as Team for exactly the same events. If the only reason you're looking at Business is SSO, price that gap carefully. You'll pay about $440 a month for SAML at that volume.
The other data types are flatter:
- ▸Logs and application metrics: 5 GB included each, then $0.50 per GB, PAYG only
- ▸Spans (tracing): 5 million included, then $0.0000016 per span reserved on Team, $0.0000032 on Business
- ▸Session replays: 50 included, then about $0.003 each, the same rate on both plans
- ▸Cron monitors: $0.78 each; uptime monitors: $1.00 each, beyond the one included
- ▸Profiling: $0.0315 per continuous profile hour, $0.25 per UI profile hour
What happens when you hit the quota
This is the part that bites teams. Sentry's quota documentation says plainly that "events and attachments that exceed your quota will not be accepted." If you haven't set a PAYG budget, a bad deploy that throws a million errors on day three means you're blind for the rest of the month. The errors you most need to see are the ones that get dropped.
Spike protection is the other half. Sentry computes a baseline from your normal volume and, when a project spikes above it, starts dropping events so you aren't charged for the surge. That protects your bill, but it also means you see a sample of the incident rather than all of it.
My setup for every Sentry account I've managed:
- Turn on spike protection for every project. It's a per-project setting, so check it again after each new service.
- Set a PAYG budget with a hard cap. A $50 to $100 cushion keeps you from going blind on a bad day, and the cap keeps a runaway loop from turning into a four-figure invoice.
- Rate-limit noisy projects at the client key (DSN) level.
- Filter before you pay. Drop known-noise errors in the SDK and use inbound filters, which Sentry says don't count against your quota. Repeated events can count, so fixing the top three noisy errors usually saves more than any plan change.
Seer, the AI add-on
Sentry's AI debugging agent, Seer, is billed separately from everything else: $40 per active contributor per month. An active contributor is anyone who makes two or more pull requests to a Seer-enabled repository in the month. On GitHub, bots marked [bot] don't count. On GitLab, every account in the group counts, including bots.
For a ten-person engineering team, that's $400 a month, which is more than Business itself at low volume. Try it on one repository before enabling it everywhere, and watch the contributor count. It resets monthly, but it grows with every repo you connect.
Self-hosting: free license, real cost
Sentry publishes a self-hosted version under the Functional Source License, which becomes Apache 2.0 after two years. You can run it anywhere, including inside a company. You just can't resell it as a competing service. Sentry describes self-hosted as "the Business plan without any software limitations and no paid tier."
The hardware requirements are the catch: at least 4 CPU cores, 16 GB of RAM plus 16 GB of swap, and 20 GB of free disk. Sentry recommends 32 GB of RAM. It runs databases, message brokers and a stack of other services on one machine through Docker Compose, and Sentry warns that it leans hard on disk I/O.
For reference, a DigitalOcean Basic Droplet with 16 GB of RAM costs $96 a month before backups. That's already more than Team plus 200,000 extra errors. The server bill isn't the real cost, though. Upgrades, database disk growth, and the night the event queue backs up are.
Self-host when data residency rules say error data can't leave your network, or when your volume is high enough that SaaS overages run into thousands a month. Otherwise, pay Sentry.
Security notes
Error reports are some of the most sensitive data your apps produce, and teams forget that. A stack trace can include request bodies, headers, cookies, user emails and, in a bad week, a database connection string.
- ▸Keep the SDK's default PII setting off (
sendDefaultPiiis false by default) unless you have a reason to turn it on. - ▸Configure server-side data scrubbing for passwords, tokens and card numbers, and test it with a fake secret.
- ▸Use SSO once you're past a handful of users. On Team, offboarding is manual. SAML and SCIM are the strongest practical reason to pay for Business.
- ▸Keep lookback short if you don't need it. Ninety days of stack traces is ninety days of exposure if an account is compromised.
My recommendation
- ▸Solo developer or side project: Developer, free.
- ▸Startup or small team: Team at $26 a month, annual. Set a PAYG cap and spike protection on day one.
- ▸Company that requires SSO: Business, but run your real error volume through the table above first. At high volume, Enterprise pricing may be worth a call.
- ▸Regulated data or very high volume: self-host, and staff it.
If you're comparing Sentry with a broader monitoring platform, our Sentry vs Datadog comparison covers the trade-offs, and the Datadog pricing breakdown shows how a per-host model compares with Sentry's per-event one.