OpenAI has been moving fast. Less than two months after GPT-5.4 dropped, the company just launched GPT-5.5. According to OpenAI, it’s the smartest model the company has shipped yet. GPT-5.5 is rolling out now to Plus, Pro, Business, and Enterprise subscribers in ChatGPT and Codex. Free users don’t get access yet.
The model’s biggest improvements land in coding, computer use, and research workflows. OpenAI says GPT-5.5 handles multi-step tasks with less back-and-forth. Instead of needing detailed instructions, the model figures out what needs to happen and works through it. Greg Brockman, OpenAI’s president, described it as a faster, sharper thinker that gets more done with fewer tokens compared to GPT-5.4.
What’s Actually New
The efficiency angle is the most interesting part for everyday users. GPT-5.5 delivers better results while using fewer tokens than its predecessor. OpenAI says the per-token speed stays the same as GPT-5.4, so responses don’t feel slower. That’s a meaningful upgrade.
On the coding side, early testers say GPT-5.5 understands why something is failing. It figures out where the fix needs to land, not just what the error message says. OpenAI’s VP of Research pointed to examples of the model cleaning up messy codebases the way a senior engineer would.
GPT-5.5 Pro, a more capable variant, goes to Pro, Business, and Enterprise users only. API access is coming soon but isn’t live yet. OpenAI says the delay is down to safety requirements for serving the model at scale. For API users, GPT-5.5 costs twice what GPT-5.4 did, at $5 per million input tokens. OpenAI argues the efficiency gains offset the higher cost for most workflows.
This is also OpenAI’s first fully retrained base model since GPT-4.5, so it’s more than an incremental update. If you’re on a paid ChatGPT plan, the upgrade is already waiting for you. At this pace, GPT-5.6 is probably already in testing.