Timothy Morano
Apr 23, 2026 18:49
OpenAI’s GPT-5.5 debuts with enhanced authorized AI capabilities, scoring 91.7% on benchmarks. Obtainable now for ChatGPT Plus and Professional customers.
OpenAI has formally unveiled GPT-5.5, its newest AI mannequin, on April 23, 2026, pushing the boundaries of synthetic intelligence in skilled and technical purposes. Early evaluations present the mannequin scoring a powerful 91.7% on Harvey.ai’s BigLaw Bench analysis suite, edging out GPT-5.4’s already sturdy 91.0%. This marks a brand new excessive for OpenAI’s generative fashions and positions GPT-5.5 as a strong instrument for authorized and technical industries.
Designed to excel in complicated and ambiguous duties, GPT-5.5 showcases developments in coding, knowledge evaluation, and autonomous activity execution. In legal-specific testing performed by Harvey.ai, the mannequin demonstrated notable enhancements in danger evaluation, deal administration, and litigation evaluation. It additionally achieved excellent scores on 43% of duties, with 87% of duties scoring above 0.80—indicating constant efficiency throughout numerous use instances. Authorized professionals have praised GPT-5.5’s capacity to supply outputs with superior organizational construction, intuitive formatting, and grounded citations.
“GPT-5.5 reveals enchancment in authorized reasoning, organizational construction, and viewers calibration,” mentioned Niko Grupen, Head of Utilized Analysis at Harvey. “These are the form of sensible beneficial properties that transfer the needle for authorized practitioners.”
Nevertheless, verbosity stays a minor problem, as some responses had been extra detailed than crucial, significantly on simple queries. Regardless of this, the mannequin’s total efficiency and flexibility signify a leap ahead in AI capabilities for skilled purposes.
OpenAI has made GPT-5.5 obtainable to paying subscribers throughout its ChatGPT and Codex platforms. Customers on Plus, Professional, Enterprise, and Enterprise tiers achieve quick entry, whereas a ‘GPT-5.5 Professional’ model is tailor-made for heavy computational workloads. This launch additionally comes with up to date API pricing, reflecting the improved performance and efficiency of the brand new mannequin.
With GPT-5.5, OpenAI is aiming to remain forward of opponents like Anthropic and Google within the race to dominate the AI house. The mannequin’s launch underscores OpenAI’s technique to place its AI as autonomous ‘digital employees,’ able to dealing with duties with minimal human intervention. Greg Brockman, President of OpenAI, said that GPT-5.5’s capacity to resolve issues with out specific step-by-step directions represents a major leap in AI autonomy.
For customers and builders, GPT-5.5 gives not simply incremental enhancements however a brand new layer of performance that would redefine workflows throughout industries. As organizations start integrating the mannequin into their operations, the implications will doubtless form market dynamics in each AI and authorized tech sectors over the approaching months.
Picture supply: Shutterstock
