Exciting News: IntelliJ IDEA Now Integrated with Codex AI Assistant

From IDEA 2024.3 the JetBrains family of IDEs shipped an AI programming assistant, but many developers reported slow responses, inability to handle complex logic, and declining plugin ratings.

On January 22 OpenAI officially opened the Codex extension, fully supporting IntelliJ IDEA, PyCharm, WebStorm, Rider and other JetBrains tools. Starting with version 2025.3 the built‑in Assistant can directly bind a ChatGPT account, raising the overall experience.

Configuration is straightforward: open the Assistant sidebar, click “Sign in Codex with ChatGPT”, install the plugin if prompted, choose “Log in via ChatGPT account”, and complete the authorization in the browser.

Beyond IDE coding, Codex can be paired with OpenClaw for Q&A, image generation, and file parsing, or used via PC and mobile clients. The daily cost is under 5 CNY, offering a cost‑effective all‑round AI helper.

The integration now runs on the latest GPT‑5.4 model and includes a GPT‑5.3‑Codex capability; on the SWE‑Bench Pro benchmark it scores 57.7 %, close to the original Codex’s 56.8 % while delivering lower latency and smoother coding assistance.

Example demonstrations include a single command to download a GitHub project (e.g., https://github.com/rymcu/mortise) even when Git is not installed, and a request for a full project overview—technology stack, module responsibilities, and quick‑start steps—where Codex automatically analyses entry points, databases, middleware, and dependencies.

When the local environment lacks required tools, Codex detects missing components, installs them (e.g., upgrades JDK 17 to JDK 21), and configures middleware such as PostgreSQL and Redis, even generating shell scripts for JDK version switching.

These scenarios illustrate that modern AI coding tools are approaching professional programmer competence.

Refactoring is a strong suit: Codex can scan an entire project, list optimization points by priority, and rewrite complex logic without altering business behavior, applying changes across related files in bulk.

Routine tasks such as generating SQL, fixing bugs, and writing README documentation are handled effortlessly.

The author, with modest front‑end skills, built a mini‑program using Codex that integrates AI‑generated avatar creation, cloud storage, membership, and payment logic; most code was generated automatically, with UI created by Bolt and fine‑tuned by Codex.

Key benefits for developers include eliminating blank‑file anxiety, rapidly understanding unfamiliar code, performing global refactoring, automatically generating and executing tests, and precise bug diagnosis within project context.

While ChatGPT remains the most frequently used tool for research, documentation, and content creation, Claude offers similar capabilities but requires a stable network and is prone to account restrictions, making ChatGPT a more reliable choice.