New Delhi: OpenAI has launched GPT-5.5, its new AI model family for ChatGPT and Codex, almost 2 months after GPT-5.4. The company says the model is built for heavier work, including coding, online research, data analysis, document creation, spreadsheets and computer use. The rollout started on April 24, (IST) for paid ChatGPT and Codex users, with API access planned “very soon.”
For regular users, this is not just another model picker update. GPT-5.5 is being pitched as a more independent assistant that can understand messy tasks, use tools, check its work and continue across steps with less hand-holding. OpenAI President Greg Brockman said GPT-5.5 is more intuitive than previous models and can do more with less human guidance, according to the context shared.
Introducing GPT-5.5
A new class of intelligence for real work and powering agents, built to understand complex goals, use tools, check its work, and carry more tasks through to completion. It marks a new way of getting computer work done.
Now available in ChatGPT and Codex. pic.twitter.com/rPLTk99ZH5
— OpenAI (@OpenAI) April 23, 2026
What is new in GPT 5.5?
OpenAI says GPT-5.5 is better at agentic work. In plain English, that means it can take a bigger job and move through more of the process on its own. Think of a developer asking it to debug a large codebase, or an office team asking it to turn rough data into a proper report. That is the area OpenAI is clearly chasing here.
The company says GPT-5.5 matches GPT-5.4’s real-world per-token latency, even with higher intelligence. It also uses fewer tokens for the same Codex tasks, which could matter for long coding and research jobs.
OpenAI has cited several benchmark scores for the model:
- 82.7 percent on Terminal-Bench 2.0
- 58.6 percent on SWE-Bench Pro
- 84.9 percent on GDPval
- 78.7 percent on OSWorld-Verified
- 98.0 percent on Tau2-bench Telecom without prompt tuning
GPT 5.5: Coding, research and office work
The biggest upgrade appears to be in coding. OpenAI calls GPT-5.5 its strongest agentic coding model so far. Early testers said it was better at holding context across large systems, finding where a bug starts and predicting what else may break after a fix.
Michael Truell, Co-founder and CEO at Cursor, said, “GPT-5.5 is noticeably smarter and more persistent than GPT-5.4, with stronger coding performance and more reliable tool use.”
The model is also being pushed for knowledge work. OpenAI says teams inside the company used Codex with GPT-5.5 to analyse speaking request data, review 24,771 K-1 tax forms across 71,637 pages, and automate weekly business reports.
GPT 5.5: Cybersecurity safeguards
OpenAI says GPT-5.5 has its strongest safeguards so far. The company has classified its cybersecurity and biological or chemical capabilities as “High” under its Preparedness Framework, but says the model did not reach the “Critical” cybersecurity level.
That means some cybersecurity requests may face stricter checks at launch. OpenAI says verified defenders can apply for Trusted Access for Cyber to reduce unnecessary refusals for legitimate defensive work.
Who gets GPT-5.5 now?
GPT-5.5 is rolling out to Plus, Pro, Business and Enterprise users in ChatGPT and Codex. GPT-5.5 Pro is rolling out to Pro, Business and Enterprise users in ChatGPT. In Codex, GPT-5.5 comes with a 400K context window for Plus, Pro, Business, Enterprise, Edu and Go plans.
For developers, OpenAI says GPT-5.5 will come to the API soon at $5 per 1 million input tokens and $30 per 1 million output tokens. GPT-5.5 Pro API pricing is listed at $30 per 1 million input tokens and $180 per 1 million output tokens.