Nick Saraev and Jack Roberts discuss demonstrations and reported benchmark resultsEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria. for Fable 5.2, including coding, computer use and visual design. Their performance comparisons are presented as the hosts’ interpretation of those results.
They compare effort levels, task costs and model pricesCost per completed task measures the total AI, tool, infrastructure, retry, and repair expense for each verified useful outcome., noting that higher effort can raise costs. They also discuss creative demos and their own plans for video workflows.
The hosts acknowledge that earlier models could produce some similar demos with more detailed instructions or stronger harnessesAn AI agent harness is the software framework that packages a model with tools, instructions, context management, execution controls, and user interaction.. Their forecasts about future models and competition remain opinions.
Watch on YouTube
