编码 agent 评分:Opus 5.5 61.7 领先 Astra 54.7
Artificial Analysis 评分显示 Opus 5.5 得 61.7、GPT-6 Astra 54.7、Sol 54.1,作者建议 Sol 干杂活、Opus 做复杂任务。

IF YOU DON’T BELIEVE OPUS 5.5 IS THE BEST CODING AGENT RIGHT NOW… LOOK AT THIS 💀 Artificial Analysis has Opus 5.5 at 61.7 👀 For comparison: Opus 5.5 → 61.7 ($11.42) GPT-6 Astra → 54.7 ($7.04) GPT-6.1 Sol → 54.1 ($1.50) But here’s the crazy part… Sol is almost at Astra’s level while costing a fraction of the price. So is Opus really worth paying ~7× more than Sol? Not always. I’d split the workload: Sol → grunt work PRs, issues, cleanup, repetitive tasks Opus 5.5 → serious work New features, complex bugs, deep/long tasks Why burn Opus tokens on boring work when Sol can handle it? 💀 Would you pay 7× more for Opus 5.5? 👀