OpenAI Launches GPT-5.5 With Self-Correction Capabilities for Complex Agentic Workflows

OpenAIOpenAI

· Updated

OpenAI released GPT-5.5 and a high-performance Pro variant across ChatGPT and Codex for paid users. The model is optimized for agentic loops, featuring the ability to verify its own outputs and handle multi-step goals with improved inference speed. This shift moves AI from a reactive assistant to a reliable partner for long-running professional tasks.

OpenAI launched GPT-5.5, a new model class designed for autonomous agent workflows and complex professional tasks. Available today for Plus, Pro, Business, and Enterprise users, the model can understand multi-step goals, utilize tools, and verify its own work. The update extends to Codex with expanded browser and computer use.

This update addresses the reliability gap in agentic AI by focusing on self-correction. While GPT-5.4 introduced computer use, GPT-5.5 optimizes inference to make high-intensity reasoning practical, following OpenAI's strategic pivot toward industrial-scale inference. It transforms the model from a reactive chatbot into an iterative research partner.

You can access GPT-5.5 now through ChatGPT or the Codex desktop app, matching efficiency gains reported by early testers. For demanding workloads, the new GPT-5.5 Pro variant offers a step change in quality for Pro, Business, and Enterprise tiers. This model is optimized for processing document context.

View the full update on openai.com
OpenAI
OpenAI
@OpenAI
X

Introducing GPT-5.5 A new class of intelligence for real work and powering agents, built to understand complex goals, use tools, check its work, and carry more tasks through to completion. It marks a new way of getting computer work done. Now available in ChatGPT and Codex. https://t.co/rPLTk99ZH5

3.5kretweets28klikes
View on X

Still wondering? A few quick answers below.

GPT-5.5 is a new class of intelligence from OpenAI designed for complex work and powering autonomous agents. It is built to understand multi-step goals, use external tools, and verify its own work to ensure tasks are completed accurately. It is currently available within the ChatGPT and Codex platforms for paid users.

GPT-5.5 is rolling out to ChatGPT Plus, Pro, Business, and Enterprise users. The more advanced GPT-5.5 Pro variant is specifically available for Pro, Business, and Enterprise tiers. These models are accessible through both the ChatGPT interface and the Codex desktop application, which is used for agentic software engineering tasks.

GPT-5.5 introduces full-stack inference improvements—the process of running a model to generate outputs—that allow for faster speeds and higher efficiency during complex tasks. Unlike earlier versions, it is specifically optimized to act as an iterative research partner that can check its own work and process document context.

GPT-5.5 is designed to power AI agents by handling complex goals and carrying more tasks through to completion without constant human direction. It features native tool use and the ability to check its own work, meaning the model can review its outputs for accuracy. This makes it more practical for demanding workflows.

Yes, GPT-5.5 is available in Codex as part of the current rollout. Codex is OpenAI platform for agentic coding and computer-based work. By integrating GPT-5.5, Codex users can leverage the model improved reasoning and ability to check its own work to manage software development lifecycles and other multi-step technical tasks more reliably.

Every HeadsUpAI update is written based on its original source and reviewed before it's published. Read our editorial standards →

OpenAI

See all AI news & updates from OpenAI →

Share this update