Fire traffic with any HTTP method at any endpoint, watch it respond in real time, and change the rate without stopping. No browser tab, no YAML, no Docker. Just a fast, native tool that does one thing well.
A real recording: the rate is pushed past what the test server can take, then brought back.
Features
Sent, Received, Failures and Error Rate scroll by smoothly while the test runs, with the target as a dashed line. Drag the slider underneath to change the rate mid-test and watch the server react.
The request rate never waits on slow responses. When the endpoint can't keep up, extra requests are shed and you see the achieved rate drop below target. That dip is the signal you want.
Max RPS, connections, queue depth, stop conditions, headers, the body and the method all apply to the running test. No restart, no lost history.
p50, p90 and p99 alongside the average, fastest and slowest request, from a histogram accurate to about 2%. A spread strip shows at a glance how long the tail is.
See how responses split across 2xx, 3xx, 4xx, 5xx and no response at all. Click any metric or response for a plain-language explanation and what to check next.
JSON, form, plain text or XML with syntax highlighting. Format and Minify in one click, and a validity check that points to the line and column of any mistake.
The exact request, and responses as cards with status, protocol (HTTP/1.1, HTTP/2 or HTTP/3), latency and size. Requests that never got a response show you why.
Stop after a number of requests, a duration, or both. Ask for 1,000 requests and exactly 1,000 are sent. Progress bars and an estimated finish time show how far along you are.
Pure Swift with zero third-party dependencies. Nothing is spawned per request, so memory stays flat at any throughput.
Swift and SwiftUI from top to bottom, not a cross-platform wrapper. Liquid Glass toolbar on macOS 26, the new tab-style switcher on macOS 27, Dark Mode, ⌘R / ⇧⌘R shortcuts, and your last setup restored on relaunch.
Under the hood
Many load tools quietly wait for slow responses, so the rate you asked for and the rate you got drift apart without telling you. Pace Maker keeps the pace and shows you exactly what the server could handle.
Screenshots
Target, load, stop conditions, headers and display settings live side by side with the chart, and every change reaches a running test. Your last configuration is restored on relaunch.
How it works
Paste a URL, pick a method and a User-Agent, add headers or a body if you need them.
Choose a target RPS and a limit: a number of requests, a duration, or both.
The chart, stats and responses update live. Drag the slider to change the rate, or pause and resume at any time.
Why Pace Maker
Great tools exist for scripted, distributed, CI-driven load. Pace Maker is for the moment you just want to hit an endpoint and see what happens.
| Capability | Pace Maker | Browser-based tools | CLI & scripting tools | Container setups |
|---|---|---|---|---|
| Native macOS app | Yes Swift & SwiftUI | No Runs in a tab | No Terminal only | No |
| Time to first test | Yes Seconds | Partly Sign-up, setup | Partly Write a script | No Images, compose files |
| Live chart while it runs | Yes | Yes | Partly Text output or add-ons | Partly Extra dashboards |
| Change settings mid-test | Yes Rate, limits, headers, body | Partly Varies | Partly Scripted stages | No Redeploy |
| No config files | Yes | Partly | No Scripts / YAML | No YAML |
| Shows shed load honestly | Yes Achieved vs. target | Partly Varies | Partly Varies | Partly Varies |
| Runs locally, no account | Yes | Partly Often cloud-hosted | Yes | Yes |
Roadmap
The engine is already protocol-agnostic. These transports are on the way.
Load test gRPC services with the same live chart.
Open and drive many concurrent WebSocket sessions.
Push plain TCP endpoints when HTTP isn't the layer you care about.
FAQ
Macs with Apple silicon running macOS 14 Sonoma or later. On macOS 26 and later you also get the Liquid Glass toolbar, and on macOS 27 the new tab-style view switcher.
You can, but be careful. Load testing can degrade or take down a service, and methods like POST, PUT, PATCH and DELETE change data on every request. Pace Maker warns you when you pick one. Use a test environment, and only test systems you own or are allowed to test.
Yes. Drag the slider under the chart to change the rate, or edit Max RPS, connections, queue depth, stop conditions, headers, the body or the method in Settings. The running test picks up every change straight away, and new headers and bodies go out with the next requests.
No. Failures are 5xx responses and requests that got no response at all, so a rate limit (429) doesn't count against the server. But when most responses are 4xx, Pace Maker says so and explains the code, e.g. a 401 means checking the Authorization header.
A token-bucket pacer releases requests at your target rate without waiting on responses. A fixed pool of workers serves them, and an adaptive connection gate sizes how many are in flight from observed latency. When the backlog is full, extra requests are shed instead of buffered forever, so the achieved RPS visibly drops when the server can't keep up.
No. Pace Maker talks only to the endpoints you point it at. There are no accounts, analytics or telemetry.
Not yet. HTTP ships today, and the engine is already protocol-agnostic. gRPC, WebSocket and raw TCP are next on the roadmap.
Native, fast and honest about the numbers. Get Pace Maker and run your first test in under a minute.