Tech • IA • Crypto
Codex just got better for developers
9.4/10TL;DR
OpenAI has integrated Codex into ChatGPT, introducing the GPT-5.6 Sol model, parallel task automation, and full-stack app deployment features aimed at streamlining developer workflows.
KEY POINTS
Codex integrated into ChatGPT
Codex is now embedded directly within ChatGPT as a dedicated workspace, marking a shift from a standalone tool to a unified development environment. The integration places coding capabilities alongside a broader “work agent,” allowing developers to manage projects, research, and execution in one interface. More than 6 million weekly users are already using Codex, reflecting significant adoption.
Launch of GPT-5.6 Sol
The new GPT-5.6 Sol model is positioned as a frontier system capable of extended reasoning on complex tasks. It introduces an Ultra mode, which increases computational depth for difficult problems but consumes more tokens. The model can autonomously break down large objectives into parallel subtasks handled by multiple agents.
Parallel multi-agent workflows
Codex now automatically distributes work across subagents, enabling simultaneous execution of multiple tasks under a single goal. Developers can monitor each agent individually or view a unified dashboard of progress. This parallelization significantly reduces time spent on large or multi-step projects.
Advanced computer and browser control
The system introduces enhanced interaction with local environments and browsers. A new “appshot” feature captures both screen visuals and application context, allowing Codex to navigate software interfaces autonomously. Demonstrations include controlling an iOS simulator to generate localized App Store screenshots without manual input.
Interactive coding and visual editing
Developers can now combine visual and code-based workflows. Codex supports point-and-click annotations within rendered apps, enabling precise change requests tied to UI elements. A built-in diff editor allows inline code modifications, supporting both manual edits and AI-assisted updates.
Built-in deployment with Sites
The expanded Sites feature enables one-step deployment of web applications. It includes hosting, authentication, databases, and file storage, allowing developers to move from concept to production without managing infrastructure. Previously limited to enterprise users, Sites is now broadly available.
Task orchestration and self-management
Codex can organize and manage its own tasks, including generating worktrees, prioritizing issues, and coordinating multiple development threads. Tasks can reference each other, creating interconnected workflows that mirror real-world software development pipelines.
Seamless context transfer from ChatGPT
Conversations and research conducted in ChatGPT can be directly imported into Codex using mentions, eliminating the need for manual copying. This creates a continuous flow from ideation to implementation within the same ecosystem.
Mobile development capabilities
Updates to the ChatGPT mobile app allow developers to manage tasks, review code changes, and connect to remote environments via SSH. Users can track branches, compare changes, and initiate work from mobile devices, supporting on-the-go development.
Integrated pull request management
Codex now supports full pull request workflows داخل the app. Developers can review changes, leave comments, resolve issues, and merge code without external platforms. The system automatically tracks branches and surfaces relevant pull requests, maintaining context throughout the process.
CONCLUSION
The integration of Codex into ChatGPT, combined with the release of GPT-5.6 Sol and expanded automation features, signals a push toward fully AI-assisted software development within a single, unified platform.