OpenAI launches GPT-5.5 for ChatGPT and Codex

OpenAI has launched GPT-5.5, a new flagship model that pushes the company further from chat and closer to execution. The model is rolling out to Plus, Pro, Business, and Enterprise users in ChatGPT and Codex, with API access coming later.

What is new

In OpenAI’s own launch materials, GPT-5.5 is framed as a model built for complex work over time, not just quick answers. It is meant to be better at understanding messy goals, using tools, checking its work, and carrying multi-step tasks through to completion. OpenAI specifically highlights agentic coding, web research, analysis, document work, and data-heavy tasks.

The company also says GPT-5.5 delivers a meaningful intelligence jump without becoming slower than GPT-5.4 in real-world serving. In OpenAI’s own evaluations, benchmarks such as Terminal-Bench 2.0, BrowseComp, and OSWorld-Verified are presented as evidence of stronger performance on tasks that require several steps and actual execution.

Why this matters

This matters because the model race is shifting away from who writes the nicest text and toward who can complete the most work with the least supervision. For executives, the implication is that value is moving from one-off prompts to operational workflows in software, analytics, and back-office work. The fact that OpenAI is prioritizing ChatGPT and Codex before the API also says something about where it sees near-term commercial value: finished work surfaces for knowledge workers and developers, not just raw model access.

Source and date validation

The original launch was published by OpenAI in the company’s own community post on April 23, 2026 at 18:09 UTC, with links to the official product page and system card. That is an official source and clearly inside the 48-hour freshness window.

Sources:

• https://community.openai.com/t/gpt-5-5-is-here-available-in-codex-and-chatgpt-today/1379630
• https://openai.com/index/introducing-gpt-5-5/
• https://openai.com/index/gpt-5-5-system-card/