OpenAI released GPT-5.5 on Thursday, calling it the company’s “smartest and most intuitive to use model” yet, with improvements across agentic coding, computer use, and scientific research.
The release comes roughly six weeks after GPT-5.4 launched in March, continuing OpenAI’s pace of near-monthly model releases. According to OpenAI co-founder and president Greg Brockman, GPT-5.5 marks another step toward a future “superapp” – a unified program that would combine ChatGPT, Codex, and an AI browser into a single service for enterprise customers.
“This model is a real step forward towards the kind of computing that we expect in the future – but it is one step, and we expect to see many in the future,” Brockman said. “It’s a faster, sharper thinker for fewer tokens compared to something like 5.4. So this means that there’s just more frontier AI available for businesses and for consumers, which is part of our goal.”
The efficiency gains are one of the headline improvements. GPT-5.5 completes tasks at the same difficulty level as its predecessor using significantly fewer tokens. OpenAI says this applies broadly – from agentic coding to document creation – with the new model also showing reduced latency on longer-form tasks.
On the coding side, OpenAI says GPT-5.5 is better at holding context across large systems, reasoning through ambiguous failures, checking its own assumptions with tools, and carrying changes through an existing codebase. For computer use, the model’s improved intent understanding translates to faster execution on complex tasks.
OpenAI chief scientist Jakub Pachocki set expectations for continued progress. “We see pretty significant improvements in the short term, extremely significant improvements in the medium term,” Pachocki said. “In fact, I would say, like, I think the last two years have been surprisingly slow.”
Mark Chen, OpenAI’s chief research officer, highlighted gains in scientific and technical work, noting the model “shows meaningful gains on scientific and technical research workflows” and could contribute to drug discovery research. Chen also said GPT-5.5 is better at navigating computer-based work than its predecessors.
The release introduces three models: GPT-5.5, GPT-5.5 Pro, and GPT-5.5 Thinking. Access to the standard GPT-5.5 starts today for ChatGPT Plus, Pro, Business, and Enterprise users, as well as Codex users. GPT-5.5 Pro is available to Pro, Business, and Enterprise tiers, but not Plus. API access is not yet available; OpenAI cited the need for “different safeguards” before that rollout proceeds.
Pricing for GPT-5.5 is higher than GPT-5.4. OpenAI points to the model’s efficiency improvements and functional gains to justify the increase.
On safety, OpenAI says the model underwent both internal and external testing. Mia Glaese of OpenAI’s technical staff said GPT-5.5 would have a significant impact on the company’s approach to deploying models for digital defense. The release also includes improved safeguards against misuse for malware generation and scam assistance.
OpenAI’s benchmark data shows GPT-5.5 outperforming both Google’s Gemini 3.1 Pro and Anthropic’s Claude Opus 4.5 across multiple evaluation categories.
GPT-5.5 is available in ChatGPT and Codex starting April 24, 2026. API availability will follow once OpenAI completes its additional safety review process.
