DeepEval Got a New Look

DeepEval just crossed 15,000 GitHub stars, and we wanted to mark the moment with something meaningful.

DeepEval started as a small PyTest integration, and today it is used by teams at companies like Google, where I used to work, Uber, the company that helps many of us get around the city, and LEGO, a brand that still feels close to home for a lot of us.

That milestone means a lot to us because it is not just a number. It reflects the developers who have tried DeepEval, shared feedback, reported bugs, contributed improvements, and helped the project keep growing.

Alongside that milestone, we also wanted to ship something meaningful on the product side: a new docs and website experience that feels easier to navigate, easier to understand, and much more developer-first.

We also wanted to take the chance to give more credit to those who make DeepEval possible. This project has always been shaped by the community, and the contributors you now see in the sidebar are a small but important way of acknowledging that DeepEval is built by many hands, not just a single team.

Whether you are getting started with your first eval, comparing frameworks, or digging into advanced evaluation workflows, the new experience is designed to make it easier to find the right path without friction and to surface the actual value DeepEval provides much more clearly.

The redesign is also meant for the agentic coding era. We wanted the docs to be more AI-agent-friendly, easier to parse, easier to search, and easier to act on, while still feeling much better for human developers. That means clearer information architecture, more approachable guides, better landing pages, and a more functional experience overall so developers can move faster from discovery to implementation.

For us, this update is more than a visual refresh. It is the beginning of a better developer experience around DeepEval, and a better way to recognize the people helping shape where it goes next.

So, what's next?

If 2025 was the year of agents, this year is definitely the year of coding agents. More and more developers are building, iterating, and shipping with AI directly inside the tools they already live in, whether that is the terminal, or their IDE of choice.

To make that possible, here are a few of the things we are working toward over the next three months:

• More features that make DeepEval more developer-friendly and more PyTest-native.
• Prompt optimization that works on entire traces
• Better local data storage and local-first workflows, so iterating with DeepEval feels easier and faster during development.
• An open-source TypeScript version of DeepEval.
• Better CLI support.

As always, we'll keep listening to feedback and contribute to any discussions, fix any bugs. Additionally, if you've ever made a PR to DeepEval, we thank you very much, If you are a user, we're glad to have you. Till next time.