Google 的 Gemini 3.0 Flash 已经发布，这是一个重大进展。它在智能水平上可与 ChatGPT 5.2、Claude Opus 4.5 等前沿模型相匹配，甚至不逊于它的“老大哥”——Gemini 3.0 Pro，但成本要低得多（便宜幅度达到数倍），同时速度快 2–5 倍。此外，它在输出时使用的 token 数量减少了约 30%，进一步提升了性价比。从昨天开始，Gemini 3.0 Flash 已成为 Google 整个生态系统中的默认模型。 Google's Gemini 3.0 Flash is out, and it's a huge deal. It matches the intelligence of frontier models like ChatGPT 5.2, Claude Opus 4.5, and even its older brother - Gemini 3.0 Pro, but it's significantly cheaper (by hundreds of percent) and 2–5x faster. It also uses 30% fewer tokens in its output, making it even more cost-effective. As of yesterday, it's the default model across Google's entire ecosystem.