GPT-5.1 in Unique AI Conduct

Overview

If your organization requires the underlying language model to be hosted in Switzerland, GPT-5.1 is currently the only option for using Unique AI Conduct.

With GPT-5.1 you get Conduct with Swiss hosting, but the trade-off is output quality: GPT-5.1 is noticeably weaker than the newer models Conduct runs on elsewhere. The limitations described on this page come from the model itself, and we are actively working to reduce them (see What we are working on below).

This page explains what to expect from Conduct on GPT-5.1, which tasks work well, and how to get the best results.

Who is it for?

Anyone using a Unique AI Conduct space that runs on GPT-5.1. Your admin configures the model and settings for your space; setup guidance is on Unique AI Conduct Space.

If you're unable to access certain features or sections of this article, it's possible that your firm doesn't have access or hasn't upgraded to the latest version. Please reach out to your internal support team for further assistance.

What to expect

GPT-5.1 does best on text-based work. The more visual or complex a task becomes, the more it struggles.

Text-based tasks work well. This includes finding information in your knowledge base and documents, researching topics on the web, using connected tools, summarizing documents, and general writing and text work.

Code-based tasks work, but need your review. GPT-5.1 does a decent job on data analysis, calculations, and simpler Excel workbooks. Results can look convincing even when they contain mistakes, so always validate the output before you rely on it. The more complex the task, the greater the risk of errors.

Visual document tasks are currently challenging. GPT-5.1 struggles with visually demanding document work such as generating PowerPoint presentations and Word documents, and editing PDFs. PowerPoint generation from company templates is particularly challenging at the moment.

Tips & Tricks

• Keep requests focused. Ask for one deliverable at a time, and split long workflows into smaller steps you can check along the way.

• Verify numbers. For any analysis or calculation, check the key figures yourself, or ask the agent to show and explain its working.

• Keep document templates simple. A small, well chosen set of template slides or pages will be easier for the model to work with compared to a massive template deck. Avoid designs requiring intricate layouts and careful positioning.

• Ask it to continue. If a result is incomplete, ask the agent to continue or fix the specific problem instead of starting over.

What we are working on

We are working to improve results on GPT-5.1 and will update this page as improvements ship. Current areas of focus include document generation using templates and improved built-in quality assurance checks.