{"id":15226,"date":"2026-06-02T11:33:28","date_gmt":"2026-06-02T10:33:28","guid":{"rendered":"https:\/\/www.clio.com\/uk\/?p=15226"},"modified":"2026-06-02T11:33:28","modified_gmt":"2026-06-02T10:33:28","slug":"gpt-5-5-agentic-legal-work","status":"publish","type":"post","link":"https:\/\/www.clio.com\/uk\/blog\/gpt-5-5-agentic-legal-work\/","title":{"rendered":"Expanding the Frontier of Legal Agentic Work: GPT-5.5 Support in Clio Work and Vincent"},"content":{"rendered":"

With the launch of OpenAI\u2019s <\/span>GPT-5.5<\/b>, we are upgrading every <\/span>Clio Work<\/span><\/a> and <\/span>Vincent<\/span><\/a> customer to the new model for agentic work and document drafting. The upgrade sets a new bar for what Clio\u2019s AI can do on substantive legal tasks, and is the foundation for the next wave of autonomous legal capability we are building across our product line.<\/span><\/p>\n

Why we’re upgrading to GPT-5.5<\/span><\/h2>\n

Over the private preview period, we evaluated GPT-5.5 through Clio\u2019s core evaluation set, designed by legal experts to reflect the work our customers do every day. This includes legal research, document analysis, drafting, <\/span>disclosure<\/span>, and scenario-based advisory. These evaluations measure end-to-end system performance, including core agentic capabilities that support longer-horizon work, such as memory management and context retention across multi-turn sessions. They assess how effectively models orchestrated through Clio\u2019s AI retrieve, integrate, and act on firm context, while leveraging our authoritative legal content library, one of the largest in the world.<\/span><\/p>\n

We evaluated GPT-5.5 against every other frontier model available to us, including OpenAI\u2019s prior-generation GPT-5.4 and models from other leading AI labs. This evaluation covered hundreds of scenarios and thousands of graded criteria, at multiple reasoning-effort levels for every model. Within Clio\u2019s AI, where models are combined with our agentic systems and legal data, GPT-5.5 delivered the strongest performance we recorded, achieving an overall score of 87.2%.<\/span><\/p>\n

\n

When powered by GPT-5.5, Clio\u2019s AI delivered the top overall benchmark score at 87.2%, higher than any other frontier model we tested.<\/b><\/p>\n<\/div>\n

Where GPT-5.5 makes the biggest difference <\/span><\/h2>\n

Two categories of work push frontier models hardest in our evaluation, and they are where GPT-5.5 moves furthest from the prior generation.<\/span><\/p>\n

The first is <\/span>legal research that requires citing the <\/b>binding authority<\/span>, including the specific case, the exact statutory section, and the leading commentary, rather than merely describing the rule. On these tasks, GPT-5.5 delivers a roughly 20% relative improvement over the prior generation, closing gaps that earlier systems consistently left open.<\/span><\/p>\n

The second is <\/span>difficult, open-ended document work<\/b>, which includes contract analysis, deal-point extraction, multi-document review, and <\/span>disclosure review across large document sets.<\/span><\/p>\n

Earlier models would reliably surface the right answer but could miss the qualifying language, scope clauses, and secondary requirements that might alter its legal meaning. GPT-5.5 reads further into the document and captures key information: the survival periods, the fraud carve-outs, the jurisdictional conditions, the conditions of exercise. Across our document-analysis scenarios, this translates to a ~7% relative improvement over the prior generation, and the difference is even larger at higher reasoning effort. The result is an answer that is not merely directionally correct but more legally complete.<\/span><\/p>\n

\n\t\n\t\t\t\n\t\t\t\t\n\t\t\n\t\t\t\t\t\t\t\t\t\n\t\t\n\t\t\t\t\n\t\t\n\t\t\t\t\t\t\t\n\t\t\t\t\t\t\n\t\t\n\t\t\t\t\t\t\t\n\t\t\t\t\t\t\t\n\t\t\t\t\n\t\t\t\t\t\n\n\n\t\n\t\t\t\t\t

More efficient use of the context window<\/span><\/h2>\n

GPT-5.5 is markedly more efficient in how it uses tokens during reasoning. It spends fewer tokens deliberating internally for the same quality of answer than other frontier models we tested. In one comparison, it used ten times fewer reasoning tokens per tool call. In practice, this means two concrete things for our customers: faster responses, and more headroom in the context window for Vincent to retain context in long, multi-turn sessions and autonomous agent work.<\/span><\/p>\n

What GPT-5.5 changes in your day-to-day work<\/span><\/h2>\n

Clio Work and Vincent\u2019s agent modes now run on GPT-5.5, and customers will start to see the difference in their day-to-day work. This includes:<\/span><\/p>\n