Performance testing upgrades

Traditional performance testing measures latency and throughput, but it doesn't tell you if your API is still behaving correctly under load. So, we've rebuilt the performance testing engine from the ground up. It now supports significantly higher load with predictable resource usage. And now you can run your existing pm.test() assertions using performance tests. Validate correctness and performance at the same time using the collections you already rely on for functional testing, so you can see exactly when and where behavior breaks as load increases.

CI support for performance testing is also in open beta. This lets you run performance tests headlessly via the Postman CLI, so you can reuse tests as release gates. Catch concurrency bugs earlier and ship APIs that behave predictably under real-world traffic.

New Errors tab

Speaking of testing, when collection runs fail, the new Errors tab speeds up debugging. Test failures appear in Failures. Unexpected errors like network issues, timeouts, or script errors appear under Errors and they're grouped by request. Jump directly to the failing request and spend less time digging through logs.

Agent Mode integrations

Agent Mode also got some updates this month. You can now create GitHub issues and Jira issues via Agent Mode with context carried over automatically.

Next up, better notifications for Slack and Microsoft Teams. You can now customize the frequency of Postman notifications in your Slack and Teams channels. Instead of getting every update, notifications now summarize new developments in your Postman workspaces. You can also customize your notifications by day of the week and time of day in a time zone you choose, so you can stay on top of API changes without overloading your channels with noise.

AI models and MCP server

Next, we're excited to introduce more flexibility with AI models. Experimenting with different models usually means switching platforms and managing separate API keys. We've redesigned AI requests to include a new provider selector and Bring Your Own Model support. Connect any OpenAI-compatible endpoint, whether it's a cloud provider, self-hosted deployment, or a local model on your machine. Test and compare models directly in Postman Collections without switching tools.

And while we're talking about AI, Postman's MCP server now includes new ways to work with APIs. Connect Postman's MCP server to Claude Desktop, Cursor, VS Code, and other MCP-compatible coding agents. The MCP server can now generate idiomatic, well-organized code from your API definitions and test your APIs by running collections directly, so your AI assistants have more context about your APIs and can help you build faster.

Team Notebooks and workflow updates

Team Notebooks lets your team collaborate on creating interactive API documentation and they're now available to use directly in Postman. Plus, you can now save time by using Agent Mode to quickly create comprehensive Notebooks. Describe what you want to build and Agent Mode will assemble a structured Notebook with guidance, runnable requests, and code snippets.

Good news for our Linux users: the Postman CLI now natively supports Linux ARM64. Run tests on ARM-based CI infrastructure without emulation for faster execution and smoother performance.

A few more improvements this month: create Jira issues directly from monitor results, sort requests alphabetically inside collections, see console logs in collection run results, and control what syncs between specs and collections in Spec Hub.

So that's what's new this month at Postman. Faster workflows, tighter integrations, and APIs that scale predictably. Give them a try and let us know what you think in the Postman community. And as always, keep it 200.