Opus 5.5 vs. GPT-6 Sol: which model won my blind taste test?

New AI models, including Anthropic’s Opus 5.5 and OpenAI’s GPT-6 Sol and Luna, offer significant improvements in speed, cost-efficiency, and agentic capabilities. Opus 5.5 integrates advanced safety guardrails while shedding the overly verbose, "annoying" personality of previous iterations, making it a viable tool for long-running research tasks. Meanwhile, GPT-6 Sol and Astra dominate in coding, personal productivity, and creative SVG generation, consistently outperforming rivals in user-centric design and readability. Despite these advancements, 3D rendering remains a persistent challenge across all frontier models, as demonstrated by the "Barbie Bench" test. Ultimately, while Opus 5.5 proves effective for complex agentic workflows, the OpenAI models remain the preferred choice for daily coding and creative tasks due to their superior ergonomics and visual output quality.