OpenAI launches GPT-5.5 — and throws the gauntlet at Anthropic

OpenAI has officially launched GPT-5.5, billed by the company as its most intelligent and user-friendly AI model to date. Unveiled on Thursday, the new release ships sweeping capability upgrades. According to OpenAI President and Co-founder Greg Brockman, GPT-5.5 marks a pivotal stride toward an era of "agentic and intuitive computing" — and brings the company significantly closer to its long-teased "super app".

A faster, sharper thinker

During a press briefing, Brockman emphasized that GPT-5.5 delivers sharper reasoning while consuming fewer tokens than its predecessor, GPT-5.4. The efficiency gain unlocks more frontier capabilities for both consumers and businesses without inflating compute costs.

Brockman and CEO Sam Altman have long championed the idea of a comprehensive "super app" — a unified platform merging the capabilities of ChatGPT, Codex, and AI-driven browsing into a single enterprise-ready service. The strategy directly mirrors that of former OpenAI co-founder Elon Musk, who is currently attempting to mold X into a similar multi-tool ecosystem.

An accelerated release cadence

The launch of GPT-5.5 follows a remarkably aggressive shipping schedule, arriving on the heels of models released last month, in December, and in November. Jakub Pachocki, OpenAI's Chief Scientist, indicated that this rapid cadence is just the beginning. Reflecting that the past two years have felt "surprisingly slow" in retrospect, Pachocki teased that both short-term and medium-term advancements will be massive.

Outperforming the competition

GPT-5.5 is engineered to excel across diverse domains:

• Enterprise workflows — automating knowledge work and executing agentic coding tasks.
• Scientific breakthroughs — assisting in advanced mathematics, experimental scientific workflows, and drug discovery — an area of massive industry growth highlighted by Chief Research Officer Mark Chen.

OpenAI also released benchmark data suggesting that GPT-5.5 consistently outperforms leading rivals, specifically calling out Google's Gemini 3.1 Pro and Anthropic's Claude Opus 4.5.

The AI arms race and security

Competitive tension with Anthropic remains a focal point in the industry. Asked how GPT-5.5 compares to Anthropic's recently announced — and controversial — cybersecurity tool Mythos, OpenAI technical staff member Mia Glaese reiterated the company's commitment to safe deployment. Glaese noted that OpenAI relies on a highly refined, durable strategy to enhance digital defense without compromising safety.

Availability

Starting Thursday, the standard GPT-5.5 model is rolling out across ChatGPT for Plus, Pro, Business, and Enterprise users. Meanwhile, the specialized GPT-5.5 Pro tier is reserved exclusively for Pro, Business, and Enterprise subscribers.