OpenAI dropped GPT-5.5 on Thursday, branding it the company’s “smartest and most intuitive to use model” yet. According to TechCrunch AI, co-founder and president Greg Brockman framed the release as another step toward OpenAI’s long-teased “super app,” a unified program that would merge ChatGPT, Codex, and the company’s AI browser into a single tool aimed at enterprise users.
This is the third model OpenAI has pushed out in roughly four months. The pace isn’t slowing. Chief scientist Jakub Pachocki told reporters that “the last two years have been surprisingly slow,” and he expects “extremely significant improvements in the medium term.” That’s a loaded statement from the person running research at the most-watched AI lab on the planet.
What’s new in GPT-5.5
- Faster thinking, fewer tokens. Brockman said the model is “a faster, sharper thinker for fewer tokens compared to something like 5.4.” Translation: more frontier capability per dollar for businesses and developers building on the API.
- Agentic coding and knowledge work. OpenAI is positioning 5.5 as a workhorse for enterprise tasks, with Mark Chen, chief research officer, noting the model is better at navigating computer work than its predecessors.
- Scientific research gains. Chen said the model “shows meaningful gains on scientific and technical research workflows” and could “help expert scientists make progress,” including in drug discovery.
- Benchmark leadership. OpenAI published data showing GPT-5.5 topping Google’s Gemini 3.1 Pro and Anthropic’s Claude Opus 4.5 across a range of evaluations. Benchmark claims from vendors always need independent verification, but this is the competitive framing OpenAI wants.
- Cybersecurity posture. Asked whether 5.5 would match Anthropic’s Mythos cybersecurity tool (which has been in the news over an unauthorized access report), OpenAI technical staff member Mia Glaese said the company has “a strong and longstanding strategy for our approach to cyber” and a “durable approach to rolling out models safely.”
The super app play
Brockman and Sam Altman have been talking about a super app for a while now. The idea: one program that handles chat, coding, browsing, and whatever else OpenAI bolts on, aimed squarely at enterprise buyers who don’t want to juggle ten different tools.
It’s also a space where Elon Musk keeps circling. Musk, an Altman rival and former OpenAI colleague, has said he wants to turn X into a super app. So you’ve got two of the most-watched founders in tech aiming at the same target, from very different starting points.
What stands out here is the framing shift. OpenAI isn’t just selling a smarter chatbot anymore. It’s selling the operating layer for knowledge work. Brockman called it “a real step forward towards the kind of computing that we expect in the future.” That’s platform language, not feature language.
Availability
GPT-5.5 is widely available starting Thursday. OpenAI says the model is rolling out to Plus, Pro, Business, and Enterprise users in ChatGPT. The heavier-duty 5.5 Pro variant is going to Pro, Business, and Enterprise tiers only.
No word on pricing changes for the underlying API in TechCrunch’s report, and free-tier ChatGPT users aren’t on the list for 5.5 access at launch.
Why it matters
Three releases in four months tells you OpenAI is running a cadence most of its rivals can’t match. Each new model tightens the pressure on Anthropic, Google, and the open-source field to keep up, and each release gets folded into the enterprise sales pitch.
The super app question is the one to watch. If OpenAI actually ships a bundled, agent-driven workspace that replaces the patchwork of AI tools most companies are duct-taping together, that’s a genuine shift in how software gets bought. If it stays a concept on press calls, then 5.5 is just another point release on the curve.
More details at the original TechCrunch AI report.