Opus 5.5 has been one of the best coding models that Anthropic has put out. As well as, Sonnet 5.5. At work, I use Opus 5.5 for orchestration and make it spin up specific subagents that use Sonnet 5.5. Both models are not chatty and almost human-like in their response and thinking.
It has been a while since I have use Cursor but I am deep in the Grok Bot and Grok Build ecosystem. Grok 4.7 is disappointing compared to Opus and Sonnet 5.5 models but when I was using Composer that model flies and it's output was consistent in my opinion.
What are you looking to accomplish with your week trial period?