OpenAI terminates the Codex product line and fully integrates it into GPT-5.5

On April 26, OpenAI announced the official termination of its dedicated programming model Codex, and integrated its core capabilities fully into the latest GPT-5.5 main model.

Confirmed by Romain Huet, Head of Developer Experience at OpenAI, the dedicated programming branch no longer exists since the GPT-5.4 version, and the GPT-5.3 released in early February this year became the last standalone Codex model. This adjustment indicates that OpenAI has once again returned to a "generalist" approach in its model architecture strategy, aiming to cover all professional scenarios, including intelligent programming, through a single powerful system.

In terms of technical performance, the integrated GPT-5.5 has shown significant improvements in autonomous handling of programming tasks (Agentic Coding) and computer resource utilization. Although the API price increased by about 20%, thanks to algorithm optimization, the GPT-5.5 consumes fewer Tokens when performing similar programming tasks compared to GPT-5.4, achieving efficient output with lower resource consumption.

Looking back, Codex had been closed for the first time in 2023 and then returned in May 2025 in the form of an o3-based architecture. Now it is integrated into the main model again, reflecting that top large models are evolving from "specialized plug-in" to "inherently versatile" when handling vertical domain tasks. This move indicates that AI programming is no longer an experimental branch requiring special maintenance but rather a fundamental indicator of the intelligence level of general-purpose large models.

Stop Using Old Instructions! OpenAI Releases the GPT-5.5 Prompt Guide: Simpler Is Better

OpenAI released GPT-5.5 prompt guidelines, urging developers to shift from lengthy instructions for older models to concise, result-oriented communication. Directly migrating old prompt stacks is counterproductive due to enhanced reasoning, making long instructions that compensated for model limitations now a performance bottleneck.....

AI Reporter or Puppet? OpenAI-Related Organization Suspected of Funding Fake News Websites

A crisis of news authenticity has emerged in the tech world: investigations reveal that the editorial team of The Wire may be composed of AI-driven fake characters rather than real people. Financial clues point to a political action committee, originating from an unusual interview request made by a journalist who claimed to be Michael Chen, but whose identity cannot be verified. Most of the website's staff "exhibit robotic characteristics."

Safe Harbor or New Peak? Singapore Is Becoming a Neutral Zone for Chinese and American AI Enterprises

Singapore is transforming from a traditional East-West gateway into a 'neutral zone' for AI, attracting global top firms to avoid geopolitical risks. For Chinese startups, it offers access to international markets with reduced political bias; for US companies, it serves as a key base to bypass talent restrictions and attract global talent, highlighting its strategic hub role in tech competition.....

OpenAI officially launches GPT-5.5, with 50 times improvement in output per megawatt

On April 24, 2026, OpenAI released its most powerful AI model, GPT-5.5, achieving a qualitative leap in intelligence and significantly improving energy efficiency. This model is designed for the era of intelligent agents, capable of autonomously planning, calling tools, and executing complex multi-step tasks. It performs exceptionally well in code generation, scientific research, and enterprise automation. The core breakthrough is shifting from passive response to active execution, significantly enhancing autonomy.

OpenAI Releases GPT-5.5: A Key Step Toward Super Applications and Autonomous Computing

On April 23, 2026, OpenAI released its flagship model GPT-5.5, which President Brockman described as the 'most intelligent and intuitive' model, bringing it closer to the vision of building an AI 'super app.' Compared to its predecessor, the model features significant improvements in speed and cognitive agility, while achieving high efficiency and low cost with optimized token usage.....