A developer collaborates with Codex, powered by GPT-5-Codex, inside an IDE where the AI assistant provides real-time coding support, reviews, and task management. Image Source: ChatGPT-5
OpenAI Launches GPT-5-Codex: A Major Upgrade to Codex for Agentic Coding
Key Takeaways: Codex and GPT-5-Codex Enhancements
OpenAI introduces GPT-5-Codex, optimized for agentic software engineering in Codex.
Codex now supports real-time collaboration across CLI, IDEs, GitHub, ChatGPT web, and iOS.
Performance boosts include a 93.7% reduction in token use for simple tasks and independent execution lasting over 7 hours for complex projects.
Code review features now rival expert human reviewers, catching critical bugs and validating correctness.
Codex CLI and IDE extensions enable image-based workflows, streamlined approvals, and integrated cloud-to-local task management.
GPT-5-Codex: A Specialized Model for Coding Agents
OpenAI has released GPT-5-Codex, a version of GPT-5 purpose-built for real-world software engineering. Unlike the general-purpose GPT-5, this model is tuned for agentic coding tasks, such as building entire projects, debugging, large-scale refactors, feature development, and code reviews. GPT-5-Codex is also more steerable, follows AGENTS.md instructions with greater accuracy, and delivers cleaner, higher-quality code—allowing developers to simply state their needs without lengthy prompts about style or formatting.
The model adapts dynamically to task complexity:
For small, interactive sessions, it delivers rapid responses.
For complex assignments, it can work independently for over 7 hours, iterating on designs, debugging, and delivering polished implementations.
OpenAI notes that GPT-5-Codex consumes 93.7% fewer tokens than GPT-5 in the simplest cases, while spending twice as long reasoning and testing on the most demanding challenges.
“GPT-5-Codex has been trained specifically for conducting code reviews and finding critical flaws,” OpenAI stated, adding, "We find that comments by GPT-5-Codex are less likely to be incorrect or unimportant, reserving more user attention for critical issues."
Expanded Capabilities: From Code Reviews to Front-End Design
GPT-5-Codex is not only more reliable for backend engineering but also demonstrates strong improvements in front-end development. The model can:
Generate polished desktop applications and mobile-ready websites, showing higher human preference ratings in evaluations.
Analyze images and screenshots provided as input, using them to validate design progress or resolve reported issues.
Produce its own visual outputs, including screenshots of completed tasks, giving developers a clearer view of its work.
While GPT-5 serves as a general-purpose AI model, GPT-5-Codex is purpose-built for agentic coding and is recommended exclusively for use in Codex environments such as the Codex CLI, IDE extension, GitHub, and Codex cloud.
Codex CLI: Community-Driven Upgrades
Since its launch in April, the Codex CLI has rapidly evolved through developer feedback. The latest update introduces several key enhancements designed to make Codex a more capable coding partner:
Image integration for wireframes, screenshots, and diagrams to help align design intent with implementation.
To-do list tracking for managing complex, multi-step coding workflows.
New tool support, including web search and MCP connections, enabling more accurate and context-aware tool use.
Three approval modes for security: read-only with explicit approvals, auto with workspace-limited access, and full access with network-enabled command execution.
Conversation state compaction, making longer coding sessions easier to manage.
A revamped terminal UI with clearer formatting for tool calls and diffs, improving readability and developer control.
Learn more about the Codex CLI in the quickstart here.
Codex IDE Extension: Seamless Local and Cloud Development
The Codex IDE extension, now available in VS Code, Cursor, and other VS Code forks, brings Codex directly into developers’ everyday workflows. The extension enables:
Faster responses by leveraging local code context, such as open files or highlighted sections.
Seamless transitions between previewing local changes and editing code, making it easier to refine and finalize implementations.
Integrated cloud task management, allowing developers to create new tasks, monitor progress, and review completed work without leaving the editor.
Cross-environment continuity, ensuring that Codex maintains context when moving between local and cloud tasks.
Learn more about the IDE extension in the quickstart.
Codex Cloud: Faster and More Integrated
Behind the scenes, the Codex Cloud has been significantly upgraded to improve both speed and usability. Key improvements include:
90% faster task completion times through container caching, reducing delays in launching new tasks or following up on existing ones.
Automated environment setup, with Codex scanning for setup scripts and installing dependencies (e.g., via
pip install) at runtime.Image-based workflows for UI design specs and bug reports, enabling developers to share screenshots and wireframes directly with the agent.
In-task browser previews and GitHub PR screenshot attachments, giving teams clear visibility into Codex’s progress and final outputs.
For more details, visit their documentation.
Code Review: Automating Critical Checks
Codex code review is designed to go beyond traditional static analysis tools by applying deeper reasoning to developer workflows. Instead of only scanning for syntax or style errors, Codex can:
Match pull request (PR) intent against diffs, ensuring that changes align with the stated purpose of the PR.
Reason through dependencies across the codebase, identifying issues that may only surface when modules interact.
Execute code and run tests directly, validating not just correctness but also functional behavior.
At OpenAI, Codex now reviews “the vast majority of PRs,” catching hundreds of issues daily—often before human review even begins. Developers can also summon Codex manually by tagging @codex review in a PR and providing additional instructions such as “review for security vulnerabilities” or “review for outdated dependencies.” This flexibility allows teams to tailor reviews for critical priorities while reducing reviewer workload.
Security and Safety: Guardrails for Agentic Coding
From the ground up, Codex is built with strong security defaults to protect codebases, data, and development environments. Key safeguards include:
Sandboxed execution by default, with network access disabled to prevent unauthorized actions or data exfiltration.
Asking for explicit permission before executing any potentially high-risk command or workflow.
Detailed transparency outputs, such as citations, terminal logs, and test results, to help developers verify Codex’s work.
Configurable controls that allow developers to fine-tune network access or set approval levels for specific environments.
In addition, Codex follows the same safety framework established for GPT-5. OpenAI has designated GPT-5-Codex as High capability in Biological and Chemical domains and has implemented robust safeguards to minimize potential misuse.
Learn more about how to securely operate and manage Codex here, and their safety approach in the system card addendum.
Pricing and Availability
Codex is included with ChatGPT Plus, Pro, Business, Edu, and Enterprise plans, with usage levels scaled to fit different types of development work:
Plus, Edu, and Business plans → Designed for individual developers and small teams, these tiers support several focused coding sessions each week. They are well-suited for classroom use, personal projects, or targeted team tasks where Codex acts as a supportive coding partner rather than a full-time agent.
Pro plan → Geared toward professional developers, Pro offers coverage for a full workweek across multiple projects. This tier supports extended daily use, allowing Codex to take on larger builds, long-running tasks, or continuous code review.
Enterprise plan → Built for organizations with distributed teams, Enterprise provides a shared credit pool that scales flexibly with demand. Teams only pay for what their developers use, making it easier to allocate resources across departments, projects, or entire engineering divisions.
Learn more about usage limits here.
For developers using the Codex CLI with an API key, direct GPT-5-Codex access will be available soon, extending the same capabilities to CLI-based workflows.
Q&A: GPT-5-Codex and Codex Updates
Q: What is GPT-5-Codex?
A: A specialized version of GPT-5 optimized for agentic coding tasks such as debugging, refactoring, and code reviews in Codex.
Q: How does GPT-5-Codex improve performance?
A: It reduces token use by 93.7% for simple tasks while working independently for 7+ hours on complex projects.
Q: Where can developers use Codex?
A: In CLI, IDEs (VS Code, Cursor), GitHub, ChatGPT web, and iOS, with seamless context transfer between local and cloud environments.
Q: What new features does Codex CLI offer?
A: Image support, improved UI, approval modes, task tracking, and external tool integration like web search and MCP.
Q: Is Codex safe for production environments?
A: Yes, Codex runs in a sandbox with limited access, requires approvals for risky actions, and provides logs and test results for verification.
What This Means: Codex as a True Teammate in Software Development
The launch of GPT-5-Codex and the accompanying upgrades to CLI, IDE, cloud, and code review tools mark a turning point for Codex. The platform is moving beyond being a coding assistant and positioning itself as a reliable engineering partner capable of managing tasks independently, collaborating across environments, and improving the quality and security of shipped software.
For developers, this means:
Faster iteration cycles, with Codex able to handle both quick prompts and long-running builds.
Reduced review overhead, as automated checks flag critical issues before human reviewers step in.
Integrated workflows, where Codex can operate seamlessly across terminal, editor, cloud, and GitHub.
Safer automation, with guardrails in place to maintain trust while scaling up AI involvement.
The trajectory is clear: Codex is becoming an embedded part of the development process, not just a tool but a teammate. With its sharper focus on real-world engineering and expanded safety framework, Codex sets the stage for more ambitious uses of AI in software development — while keeping humans firmly in control.
Editor’s Note: This article was created by Alicia Shapiro, CMO of AiNews.com, with writing, image, and idea-generation support from ChatGPT, an AI assistant. However, the final perspective and editorial choices are solely Alicia Shapiro’s. Special thanks to ChatGPT for assistance with research and editorial support in crafting this article.