OpenAI Launches GPT-5.5: Enhanced Speed, Coding, and Research

OpenAI has announced the release of GPT-5.5, its latest and most advanced AI model, designed to deliver faster performance and enhanced capability for complex tasks like coding, research, and data analysis. The model is rolling out immediately for ChatGPT Plus, Pro, Business, and Enterprise users, with API access expected soon.

GPT-5.5 introduces significant improvements over its predecessor, GPT-5.4, particularly in its ability to handle complex, multi-step operations autonomously. According to OpenAI, the new model achieves higher accuracy across key benchmarks, such as an 82.7% success rate on Terminal-Bench 2.0 (which tests for multi-step command-line tasks), compared to 75.1% achieved by GPT-5.4. Additionally, it delivers state-of-the-art performance in coding benchmarks like Expert-SWE, which evaluates long-horizon coding tasks.

Efficiency Gains Without Latency Trade-offs

One of GPT-5.5’s standout features is its ability to maintain GPT-5.4’s real-world serving latency while being more token-efficient. This is particularly relevant for high-volume users in coding and research, where efficiency translates directly to cost savings. For example, GPT-5.5 reportedly uses significantly fewer tokens to complete tasks previously handled by OpenAI’s Codex.

"Builders want better outcomes, not endless iteration," said Fabian Hedin, CTO of Lovable, an early tester of the model. He highlighted how GPT-5.5's ability to land complex tasks in fewer attempts transforms how users think about AI-assisted development.

Targeting Key Use Cases: Coding, Knowledge Work, and Research

The model’s strongest gains appear in coding and agentic workflows, where it excels at debugging, testing, and even long-term planning across toolchains. In knowledge work, GPT-5.5 has improved in generating documents, spreadsheets, and presentations, with alpha testers noting its ability to turn "messy business inputs" into actionable plans. OpenAI’s internal teams have already utilized GPT-5.5 in workflows ranging from tax form analysis to automating weekly business reporting, saving significant time.

In scientific research, GPT-5.5 outperformed GPT-5.4 on benchmarks like GeneBench, aimed at multi-stage data analysis in genetics and quantitative biology. One early tester, Derya Unutmaz, an immunology professor, used GPT-5.5 to analyze a gene-expression dataset with 28,000 genes. "What would have taken months was completed in a fraction of the time," he noted.

Safety and Cybersecurity Enhancements

OpenAI emphasized the enhanced safety features of GPT-5.5, particularly in the context of cybersecurity. The model underwent rigorous testing with internal and external red teams, focusing on advanced capabilities in biology and cybersecurity. As part of its safeguards, OpenAI has deployed stricter classifiers to ensure responsible use, while making expanded access available for verified cybersecurity professionals through a "Trusted Access" program.

Pricing and Availability

GPT-5.5 is now live for ChatGPT Plus, Pro, Business, and Enterprise subscriptions. Pricing details for API developers include $5 per 1M input tokens and $30 per 1M output tokens. Additionally, GPT-5.5 Pro offers even higher accuracy and is priced at $30 per 1M input tokens and $180 per 1M output tokens.

While GPT-5.5 comes at a premium over GPT-5.4, users are expected to benefit from its enhanced token efficiency, which reduces overall operational costs for many use cases.

What’s Next?

OpenAI plans to expand GPT-5.5’s availability to API users in the coming weeks, opening up opportunities for developers to integrate the model into custom applications and workflows. With its focus on efficiency, broad capabilities, and safety, GPT-5.5 positions OpenAI as a leader in the next phase of AI-driven productivity.