Svix pricing in 2026: the real 30-day webhook bill (and why $490 is almost never a volume decision)

Quick answer: Svix bills webhooks on "attempted messages," and in 2026 both the Free plan and the $490 Professional plan include the same 50,000 messages a month. Free is $0 with overage at $0.0001 per message; Professional buys throughput (200 to 800 messages per second) and features like static source IPs, not more volume. On volume alone, Free-plus-overage stays cheaper until roughly 4.95 million messages a month. The two line items that actually surprise people are the 64 KiB chunking rule and a feature gate that has nothing to do with how many webhooks you send.

What the meter actually counts

Svix sends webhooks on your behalf and bills a single unit: the attempted message. Two things are explicitly free on Svix's 2026 pricing page: retries, and messages that get filtered because no endpoint is subscribed. So a flaky consumer that forces five delivery attempts still costs you one message, and an event with no listeners costs you nothing.

The part nobody reads is the chunk rule. Svix counts each message in 64 KiB blocks. A payload up to 64 KiB is one message. A 180 KiB payload is three. That single line turns "50,000 free messages" into a moving target that depends on how fat your JSON is, not how many events you fire.

The three plans, in real 2026 numbers

Read that included-messages column twice. Free and Professional both include 50,000. You do not buy volume when you jump to $490. You buy 4x throughput and a short list of features.

Three real 30-day bills

Same rough event count, three very different invoices, each for a reason that is not obvious from the sticker.

Cohort 1: the disciplined SaaS pays $0

A B2B app fires order.created, order.updated, and subscription events: about 1,200 events a day, 36,000 a month. Each webhook is a thin envelope, well under 64 KiB. Average throughput is under half a message per second. That is 36,000 billable messages against a 50,000 free allowance. The bill is zero, and it stays zero until either the event count or the payload size moves.

Cohort 2: the payload-bloat app triples its own meter

Same 1,200 events a day. But this team ships the entire nested order object on every webhook, plus a base64 receipt thumbnail, so each payload lands around 180 KiB. At three 64 KiB chunks per message, 36,000 events become 108,000 billable messages. Overage is 58,000 messages at $0.0001, so $5.80 a month.

The $5.80 is not the point. The point is that they budgeted for 50,000 free events and actually got about 16,600, because every event spends three units. They hit the paywall at one third of the volume they planned for, and the invoice never explains why.

Cohort 3: the static-IP requirement is a $490 tax on 21,000 messages

A fintech customer will only accept your webhooks if they arrive from a fixed set of IP addresses they can allowlist in their firewall. Static source IPs are a Professional feature. This app sends 700 thin events a day, 21,000 a month, comfortably inside the free tier and worth exactly $0 on volume. One compliance checkbox moves it to $490 a month, or $5,880 a year, none of which is about webhook count. A nightly batch that bursts above 200 messages per second produces the same jump for a different reason.

The $490 cliff is a feature decision, not a volume one

Because Free carries the same overage rate as Professional, you can run enormous volume on Free before the plan price wins. Solve for it: $490 divided by $0.0001 is 4.9 million overage messages, plus the 50,000 included, so about 4.95 million messages a month before Professional beats Free on volume alone. At 4.95 million messages spread across a month that is under two messages per second on average, nowhere near the 200 msg/s Free ceiling.

So the honest decision rule is short. Upgrade for burst throughput above 200 msg/s, for static source IPs, or for the unbranded experience. Do not upgrade because you crossed 50,000 messages. Below roughly 5 million a month, staying on Free and paying overage is cheaper.

The cheaper lever: trim the envelope

The single highest-leverage move is to stop shipping fat payloads. Send a thin event: the type, the resource ID, and a timestamp, and let the consumer call your API for the full object if it wants detail. That keeps every message inside one 64 KiB chunk, which turns Cohort 2's 108,000 billable messages back into 36,000 and restores the full 50,000-event free headroom. It also tightens your webhook contract, since consumers stop depending on the exact shape of a payload you might change later. This is the same "match the unit you are billed on" discipline that shows up in Ably's message meter, which we broke down in the Ably pricing teardown.

The escape hatch: Svix is MIT open source

If a feature gate rather than volume is what pushes you to $490, self-hosting is on the table. The Svix server is open source under the MIT license as svix-webhooks, a Rust service backed by PostgreSQL and an optional Redis 6.2+ for the task queue. Run it yourself and both the per-message meter and the throughput ceiling disappear, and you control the egress IPs and branding outright, which are the exact two things the paid plan was gating.

The crossover is about features, not compute. Professional is $5,880 a year. A modest self-host stack (a small VPS, managed Postgres, a small Redis) runs in the rough range of $30 to $60 a month in 2026, call it $360 to $720 a year, plus your own operations time. If you were only paying $490 for static IPs or an unbranded experience, self-hosting delivers both for a fraction of the cash cost. The trade you accept is real: you now own uptime, upgrades, and the managed add-ons and SLAs you give up. When the question is genuinely build versus buy, we walk the full math in when DIY beats SaaS at scale.

When paying actually earns its price

Managed Svix is worth the money in three honest cases. First, you send real burst volume and cannot afford to be throttled at 200 msg/s. Second, you have enterprise customers whose security teams demand static source IPs or a signed SLA, and self-hosting those guarantees would cost more in engineering time than $490 a month. Third, you want the surrounding products, since competitors such as Hookdeck meter events and throughput on a different curve and the right pick depends on which shape fits your traffic. Outside those cases, the Free plan with disciplined payloads is the correct default, and the $490 line item is a tax you have not yet earned.

Math check: Svix's $490 Professional plan includes the same 50,000 messages as the $0 Free plan; on volume alone, Free-plus-overage stays cheaper until about 4.95 million messages a month, so the upgrade is a throughput-and-features decision, not a webhook-count one.