The latest AI model from OpenAI arrives weeks after Anthropic’s Claude Mythos shook the industry
OpenAI is not slowing down. The company announced Thursday the launch of GPT-5.5, its latest artificial intelligence model, describing it as a meaningful step forward in the areas of coding, computer operation and deep research capabilities — and rolling it out immediately to paying subscribers.
The announcement arrives less than two months after OpenAI released GPT-5.4, a timeline that underscores just how rapidly the AI sector is moving and how intense the competition among the industry’s biggest players has become.
What GPT-5.5 can actually do
According to OpenAI, GPT-5.5 is designed to handle complex, ambiguous tasks with less human guidance than previous models required. The company says it excels across 5 specific areas: 1. analyzing data, 2. writing and debugging code, 3. operating software directly, 4. researching online, and 5. creating documents and spreadsheets autonomously.
OpenAI President Greg Brockman described the model during a briefing with reporters as a genuine shift in how people might interact with computers going forward. He framed it as a model that can look at an unclear problem and work out what needs to happen next — without needing someone to spell out every step.
That kind of autonomous problem-solving capability is increasingly becoming the defining characteristic that separates newer AI models from their predecessors, and it appears to be the quality OpenAI is most eager to highlight with this release.
The cybersecurity question
Not everything about GPT-5.5 is straightforward. OpenAI confirmed that the model meets the criteria for its internal “High” risk classification — meaning it could amplify existing pathways to severe harm in cybersecurity contexts. The company was careful to note, however, that it does not cross the higher “Critical” threshold, which would indicate the potential for entirely new and unprecedented pathways to harm.
To address those concerns, OpenAI said GPT-5.5 underwent extensive third-party safeguard testing and red teaming focused specifically on cyber and biological risks. The company’s vice president of research, Mia Glaese, noted during the Thursday briefing that OpenAI has been iterating on its cybersecurity safeguards for months as its models have grown increasingly capable in that area.
The topic carries particular weight right now. Earlier this month, Anthropic unveiled its Claude Mythos Preview model, which immediately drew significant attention from both Wall Street and government officials due to its advanced ability to identify software vulnerabilities and security flaws. Anthropic opted to limit Mythos’ rollout specifically because of those capabilities, and the industry has been grappling with the implications ever since.
OpenAI’s decision to proceed with a broader GPT-5.5 rollout while flagging the High risk classification will likely invite scrutiny from the same corners that have been closely watching Anthropic’s approach.
Who gets access and when
GPT-5.5 is available starting Thursday to OpenAI’s paid subscriber base across 4 tiers: Plus, Pro, Business and Enterprise users. Access is live in both ChatGPT and Codex, the company’s coding-focused assistant.
OpenAI said the model will also become available through its application programming interface in the near future, though the company noted that API deployments require different safeguards to be in place before that rollout can proceed.
A race with no finish line
The pace of releases from OpenAI this year reflects the broader reality of an industry where the competitive pressure to ship newer and more capable models has become relentless. Google remains a formidable rival, and Anthropic’s Mythos Preview has clearly energized investors and raised the bar for what the market expects from frontier AI models.
GPT-5.5 is OpenAI’s answer to that pressure — a model the company believes sets a new foundation for how artificial intelligence will interact with computers and assist with knowledge work going forward. Whether it holds that position for long is, given recent history, very much an open question.
Source: CNBC (Ashley Capoot)