Arena 实测对比：Gemini 4 Pro 与 Opus 5.5 同处第一梯队

0xAI42.exe @0xAI42exe原推发布于 2026.09.28

在 X 上打开原推

Spent some time with Gemini 4 Pro (arena), Gemini 3.8 Flash, Claude Opus 5.5 and GPT-6 Astra side by side. My read, not a benchmark table. Gemini 4 Pro and Opus 5.5 are sitting in a different tier right now. Both hold the thread on long tasks and stop inventing things when the question gets narrow. 3.8 Flash surprised me for what it is. For a fast tier it holds up far better than the price would suggest. Astra lands in the middle. I would put it above 3.8 Flash, and behind Gemini 4 Pro and Opus 5.5 on the work I care about. Treat this as one person's impression. Your workload will rank them differently.