--- title: "Performance Testing | Postman API Platform" description: "Load test your APIs under real-world traffic. Ramp up virtual users and find your API's breaking point before your users do." url: https://www.postman.com/product/performance-testing/ --- # Performance Testing | Postman API Platform > Load test your APIs under real-world traffic. Ramp up virtual users and find your API's breaking point before your users do. ## Performance Testing Load test your APIs with real-world traffic. Ramp to millions of virtual users in the cloud and see exactly where they break. [Read Docs](https://learning.postman.com/docs/tests-and-scripts/performance-testing/performance-test-configuration) --- ## Simulate production traffic and find where your APIs break The collection you built to design and functionally test your API is also your load test. No new scripts, no separate load-testing tool, no new language to learn: the tests you already wrote are the load test. ### Load test your API Point Collection Runner at a collection and generate load, locally or on Postman's cloud infrastructure. Postman ramps up virtual users, each looping through your requests in parallel and running your scripts and tests as it goes. ### Simulate real traffic with load profiles Choose Fixed, Ramp up, Spike, or Peak. Set an initial load and a maximum, and shape how virtual users arrive over the test. ### Run performance tests on the CLI Run load tests from the Postman CLI and gate the build on success criteria like response time and error rates. When a run misses the bar, the build fails and the regression never ships. --- ## Describe the load and let Agent mode run it Agent mode configures the load profile, runs your collection under increasing virtual users, tracks p95 latency and error rate at each step, and tells you the exact load where your API starts to break. --- *CAPABILITIES* ### Build resilient APIs with performance testing Turn a collection you already have into a load test, scale it in the cloud, and see which tests break under the load. Every metric updates live, and the full report is a link your team can open. ### Scale past the limits of a single machine Cloud runs execute on Postman's managed infrastructure and spread the load across multiple machines, so your virtual user count isn't capped by your own system resources. Push into the millions to simulate the high-volume traffic your APIs see at peak, then close your computer and let the run finish without you. ### Run your functional tests as part of every load test Your collection's pm.test() assertions keep running as virtual users climb. The Tests vs VU heatmap shows exactly which ones start failing and at what load, while the error list surfaces the requests behind them. Functional results and load metrics live in the same run, not two separate tools. ### Data-driven runs powered by Datasets Connect a MySQL or Postgres database, a CSV, or several sources at once, then shape the rows with SQL views. Every virtual user pulls its own record from a live source, so a run at scale sends realistic, varied traffic instead of replaying one payload. ### Performance tests on the CLI Run the same load test from the Postman CLI and set a success criterion like p95 under 500 ms with the --pass-if flag. When your API misses it, the run fails and the build stops, so regressions never ship. ### Shape the load with profiles Choose Fixed, Ramp up, Spike, or Peak, set an initial and maximum load, and model the traffic pattern you expect in production. ### Setup and teardown at run level Sign in, seed data, and clean up once per run instead of once per virtual user. Setup and teardown are collections, so you version and reuse them like everything else you build in Postman. --- *RESOURCES* ## Go deeper on performance testing ### Configure a performance test Set virtual users, test duration, and a load profile, then run your collection locally or in the CLI. [Read docs](https://learning.postman.com/docs/tests-and-scripts/performance-testing/performance-test-configuration) ### Read the metrics What throughput, average response time, P90/P95/P99, error %, and failure % each tell you about your API under load. [Read guide](https://learning.postman.com/docs/tests-and-scripts/performance-testing/performance-test-metrics) ### Assert under load How your collection's assertions run as virtual users increase, and how the Tests vs VU view surfaces where your API breaks. [Read docs](https://learning.postman.com/docs/tests-and-scripts/performance-testing/test-assertions/overview) --- ## See performance testing on your own APIs Chat with a Postman expert to see how teams load test their APIs by ramping virtual users, reading the metrics live, and finding the breaking point before their users do. --- ## Broken collaboration leads to broken APIs. Ninety-three percent of API teams still face collaboration blockers. The 2025 State of the API report reveals how you can unlock the productivity gains that API-first promises, but scattered tooling prevents. [Read the Report](/state-of-api/2025/)