Pat Simmons reviews reported Opus 4.8 improvements, including model effort controls, faster inference and a research preview for parallel workflows. He treats benchmark claims and early impressions as provisional.
He gives Opus 4.7, Opus 4.8 and GPT 5.5 the same prompt to build a flight simulatorEvaluation is the systematic assessment of a system's behaviour or results against explicit criteria.. He observes the time spent checking and revising and then tries each result.
Simmons says the Opus 4.8 build had the most appealing controls and rendering in his demonstration, while noting that a single test does not settle broader performance.
Watch on YouTube




