Zed is built by people who previously built Atom, and the lesson they appear to have taken from that experience is that editor performance matters more than feature count. It is written in Rust, it starts instantly, and it stays responsive in large files where other editors struggle. The documentation is concise and direct, covering configuration, collaboration and the AI features without the padding that afflicts most product documentation. Performance is the honest differentiator and the docs are appropriately matter of fact about it.
Rather than claiming to be fast, the material explains the architectural decisions that make it fast, including the rendering approach and the concurrency model. For a developer who has watched an editor stutter on a large file, the difference is immediately obvious and does not need marketing. The AI integration model is the part I most approve of. Rather than bundling a subscription, you bring your own API key from whichever provider you prefer, which means you control the cost, you know exactly what you are paying, and you can switch providers as the models change.
The documentation explains the setup for the main providers clearly. That model respects the user in a way that bundled subscriptions do not, and it is a meaningful difference from the competition. The agent panel documentation covers the assisted editing features properly, including how context is gathered and what gets sent. Being open source means those claims are verifiable rather than trusted, which for anyone with security concerns about AI coding tools is a genuine advantage that no proprietary competitor can match.
The feature set is less mature than dedicated AI editors and the docs do not oversell it, which I respect. Configuration is text based and well documented, which suits the audience. Settings live in a file you can version control and share, keybindings are explicit, and the documentation covers the options systematically. Anyone who has fought with a graphical settings interface will appreciate the approach.
The ecosystem gap is the real cost and no amount of good documentation closes it. Extensions, language servers, debuggers and the enormous accumulated ecosystem around the incumbent editor represent years of community work, and this is not going to match that soon. The documentation is honest about what is supported and less clear about what is missing, and anyone with an unusual language or a dependency on a specific extension should verify before switching rather than discovering the gap mid project. Documentation lag is visible in places.
The product moves quickly, particularly on the AI features, and some pages describe behaviour that has since changed. That is the normal cost of a fast moving open source project and it does mean checking the current behaviour when something does not match. Language support quality varies more than the documentation signals. The major languages are well served and the long tail is inconsistent, and a page listing support maturity by language would save people a lot of trial and error.
Three point nine. Clear, honest documentation for an editor with a genuine performance advantage and the most respectful approach to AI integration in the category. Held back only by an ecosystem that cannot yet match the incumbent, which the documentation acknowledges more readily than most products would.