pgadmin.org

Use the desktop application if you are the only person who needs to reach the database and you want the shortest path from install to query. Use the web application if several people need access, or if you want one centrally maintained pgAdmin that any browser can reach. Both modes run the same pgAdmin 4 code and expose the same object management and query tooling, so the decision is about access and operations, not features.

The two deployment modes

pgAdmin is the graphical management tool for PostgreSQL. According to the project site, it is cross-platform: you can "run it on Windows, macOS or Linux as a desktop application, or deploy it as a web application reachable from any browser."

What stays the same in both modes

The choice does not change what you can do. The site describes the same capabilities for pgAdmin generally:

• Complete object management — "Create, browse and maintain every PostgreSQL object through a graphical interface that keeps the generated SQL in plain sight."
• A serious query tool — "Syntax highlighting, auto-completion, query history and a graphical EXPLAIN for working out where the time actually goes."

So if your only concern is whether the web version is a stripped-down edition, it is not presented that way. The feature set is the tool's feature set.

How to decide

Choose the desktop application when:

• You are the primary or only user of the databases you administer.
• You want to install and start working without standing up a server.
• Your databases are reachable from your own machine.

Choose the web application when:

• Multiple people need to connect, and you would rather not install and update pgAdmin on every machine.
• You need access from browsers on devices where you cannot install software.
• You want a single deployment to manage and upgrade.

A note on versions

pgAdmin 4 is actively released. The site's news archive lists, for example, pgAdmin 4 v9.18 (Sept. 17, 2026), described as including 29 bug fixes and new features, including fixes for four security vulnerabilities. If you run the web application, that release cadence is an argument in its favor: you patch one deployment instead of chasing every desktop install.

pgAdmin 4 has added AI functionality in three areas, documented as a three-part series on the official blog by Dave Page: LLM configuration and reports, an AI chat agent, and AI insights for EXPLAIN plans. The features are described as "coming in pgAdmin 4" in the first post of the series, so treat them as a capability set introduced together rather than as long-standing, stable tooling. If you want to use them, plan on configuring an LLM connection first — the blog series explicitly starts there.

The three AI features

1. Configuration and reports

The first post in the series (March 9, 2026) covers LLM configuration and reports. This is the setup layer: you tell pgAdmin which language model to talk to, and the reports feature builds on that connection. Nothing else in the AI feature set works until this is in place.

2. AI chat agent

The second post (March 10, 2026) introduces an AI chat agent inside pgAdmin. Rather than switching to a separate chat window or browser tab, the agent lives in the tool you're already using to manage PostgreSQL objects — which matters if your workflow is "inspect object, ask question, adjust query" without leaving the app.

3. AI insights for EXPLAIN plans

The third post (March 16, 2026) applies AI to query plans. This is the most concretely useful of the three for day-to-day tuning: pgAdmin already has a graphical EXPLAIN, and the AI layer adds interpretation on top of it. Instead of reading a plan tree yourself and reasoning about where time goes, you get an explanation of the plan.

How the features fit together

The dependency is one-directional: configuration is the gate, and the other three are consumers of it.

What to check before relying on these features

• Version. The AI posts are dated March 2026 and describe functionality "coming in pgAdmin 4." Check the release notes for your installed version rather than assuming the features are present. The most recent release noted on the site is pgAdmin 4 v9.18 (Sept. 17, 2026), which lists 29 bug fixes and new features but does not enumerate AI changes in the excerpt available.
• LLM access. The configuration step implies you supply or point to a language model. The site material doesn't state which providers are supported or whether any are bundled, so verify that against the configuration post before planning a rollout.
• Sending plans and schema to a model. AI insights for EXPLAIN and a chat agent both mean query text, plans, and possibly schema details leave your pgAdmin instance for a model endpoint. If that's a problem for your data, the AI features may not be usable regardless of how well they work.

Where to read the details

The three posts are the primary source and are meant to be read in order:

1. "AI Features in pgAdmin: Configuration and Reports" — March 9, 2026
2. "AI Features in pgAdmin: The AI Chat Agent" — March 10, 2026
3. "AI Features in pgAdmin: AI Insights for EXPLAIN Plans" — March 16, 2026

All three are by Dave Page on the pgAdmin blog. Start with the first one; the later posts assume the configuration described there.

pgAdmin's query tool is a graphical SQL editor built for writing, running, and tuning queries against PostgreSQL. According to the pgAdmin project site, it provides syntax highlighting, auto-completion, query history, and a graphical EXPLAIN for working out where query time actually goes. If you write SQL by hand and want to see both the query and its execution plan without leaving the client, this is the part of pgAdmin you will use most.

Writing SQL in the query tool

The editor is designed to keep SQL readable while you type:

• Syntax highlighting — keywords, strings, and other SQL elements are visually distinguished, which makes long statements easier to scan.
• Auto-completion — as you type object and keyword names, the tool suggests completions, reducing typos in table, column, and function names.
• Generated SQL in plain sight — pgAdmin's object management keeps the SQL it generates visible, so actions you take through the graphical interface can be inspected as SQL rather than treated as a black box.

That last point matters when you are learning: you can perform an operation through menus and then read the equivalent statement, or copy it into the query tool and adapt it.

Running queries and reviewing history

When you execute a statement, the query tool returns results in a grid and records the execution in query history. History lets you:

• Recover a statement you ran earlier without retyping it.
• Compare variations of a query you are iterating on.
• Trace what was actually executed during a session.

This is useful for tuning work, where you typically run the same query several times with small changes and need to keep track of which version produced which result.

Tuning with graphical EXPLAIN

The headline tuning feature is the graphical EXPLAIN. Instead of reading a raw text plan, you get a visual representation of the plan tree, which helps you locate where time is being spent — for example, which node in the plan dominates execution.

The practical workflow is:

1. Write or paste the query into the query tool.
2. Run EXPLAIN (graphically) rather than only executing the query.
3. Inspect the plan to find the expensive operation.
4. Adjust the query — or the indexes and statistics it depends on — and re-run EXPLAIN to compare.

The site also notes newer AI-assisted functionality around EXPLAIN plans: a March 2026 blog series by Dave Page covers AI Insights for EXPLAIN Plans, an AI Chat Agent, and LLM configuration and reports. Those posts are the place to look if you want the assistant to help interpret a plan.

What the query tool is not

The query tool is a client for PostgreSQL, not a database engine or a standalone SQL dialect. It connects to a PostgreSQL server, and the SQL you write is executed there, so performance characteristics come from the server, the schema, and the data — the tool helps you see and iterate, it does not make a slow query fast by itself.

Choosing based on how you work

pgAdmin runs as a desktop application on Windows, macOS, or Linux, or as a web application reachable from a browser, so the query tool is available in either deployment. For the current release details, the project announced pgAdmin 4 v9.18 on Sept. 17, 2026, with 29 bug fixes and new features, including four security fixes (CVE-2026-86861 through CVE-2026-86864); release notes on the site list the specifics.

pgAdmin 4 ships as numbered releases on an ongoing cadence, and each release bundles bug fixes, new features, and — when needed — security patches. The project publishes a release announcement plus detailed release notes for every version, so you can track exactly what changed before you upgrade. The most recent example in the project's news feed is pgAdmin 4 v9.18, announced September 17, 2026, which included 29 bug fixes and new features along with fixes for four security vulnerabilities (CVE-2026-86861 through CVE-2026-86864).

How releases are structured

Each version announcement follows a consistent pattern:

• A version number in the form v9.x (for example, v9.18), indicating a point release on the current major line.
• A count of changes, such as "29 bug fixes and new features."
• Security disclosures, listed by CVE identifier when the release addresses vulnerabilities.
• A "notable changes" section that highlights the most user-visible items, split into Features and other categories.
• A pointer to the full release notes for the complete change list.

The announcement itself is a summary; the release notes are where the exhaustive detail lives. If you need to know whether a specific bug or feature is in a given build, the release notes are the authoritative source.

What a release can contain

Using v9.18 as a concrete example, the notable changes spanned both usability and security:

Features

• Collapse and restore the Object Explorer by re-clicking the current workspace icon, similar to the VS Code sidebar, with the choice remembered across refreshes.
• A keyboard shortcut (Ctrl+Alt+B by default) that performs the same toggle, configurable through a new toggle_object_explorer preference.

Security hardening

• The default Content-Security-Policy was tightened so inline scripts run under a per-request nonce rather than a blanket 'unsafe-inline'.

This mix is typical: interface refinements and preference additions sit alongside security work in the same release.

How to track updates

Two project pages carry the update trail:

The news archive keeps past announcements, so you can review the release history rather than only the latest version. The website also links to the release notes directly from each announcement.

Why the cadence matters for you

Because fixes and security patches arrive together in numbered releases, the practical takeaway is:

• If you maintain a pgAdmin installation, check the announcement for the version you're running and read the security section — CVE fixes are called out explicitly.
• If you're evaluating a feature, confirm it appears in the "notable changes" or release notes for the version you plan to install, rather than assuming it exists in all builds.
• If you need the complete picture, go past the announcement to the release notes; the announcement intentionally summarizes.

The project describes pgAdmin as the leading open-source graphical management tool for PostgreSQL, and this release-and-notes structure is how that tool's changes reach users.

pgAdmin is the leading open-source graphical management tool for PostgreSQL. You use it to create, browse, and maintain PostgreSQL objects through a GUI instead of typing every command by hand — and it keeps the SQL it generates visible, so you can see exactly what is being run against your database. It runs as a desktop application on Windows, macOS, or Linux, or as a web application reachable from any browser.

What pgAdmin actually does

The tool is organized around a few core jobs:

• Object management — create, browse, and maintain every PostgreSQL object through a graphical interface. The generated SQL stays in plain sight, so the GUI doubles as a way to learn or verify the underlying statements.
• Query work — a query tool with syntax highlighting, auto-completion, query history, and a graphical EXPLAIN for working out where query time actually goes.
• Cross-platform access — the same tool is available as a desktop app or as a browser-accessible web deployment, so teams can standardize on one interface across operating systems.

Who it is for

pgAdmin fits anyone who works with PostgreSQL and wants a visual layer over the database: developers writing and debugging queries, DBAs managing objects and permissions, and teams that need a shared, browser-reachable management interface rather than per-machine desktop installs.

If you only ever run a handful of fixed queries from a script, a GUI adds little. If you regularly inspect schemas, trace slow queries, or manage objects across environments, the graphical EXPLAIN and object browser are the parts that earn their place.

Recent development direction

pgAdmin 4 continues to ship regular releases. Version 9.18 (announced Sept. 17, 2026) included 29 bug fixes and new features, among them four security vulnerability fixes (CVE-2026-86861 through CVE-2026-86864). Notable changes in that release:

• Collapse and restore the Object Explorer by re-clicking the current workspace icon, in the manner of the VS Code sidebar, with the choice remembered across refreshes. A keyboard shortcut (Ctrl+Alt+B by default) does the same and can be changed through the toggle_object_explorer preference.
• A hardened default Content-Security-Policy so inline scripts run under a per-request nonce rather than a blanket unsafe-inline.

The project has also been adding AI functionality to pgAdmin 4, covered in a three-part blog series by Dave Page: LLM configuration and reports, an AI chat agent, and AI insights for EXPLAIN plans.

Getting oriented

If you are new to the tool, the official video library is a practical starting point — for example, walkthroughs on using the Restore dialog, authenticating pgAdmin 4 with GitHub, and setting up Kerberos and Active Directory. These cover the setup and operational tasks that are least obvious from the interface alone.