返回全部动态

谷歌发布Gemini 3.7 Flash,编码性能提升且价格减半

原标题:Gemini 3.7 Flash lands with coding gains and undercuts its three-week-old predecessor's price by 50%

THE DECODER模型发布质量 72

AI 摘要

谷歌发布了Gemini 3.7 Flash模型,距离上一代Gemini 3.6 Flash仅三周。该模型在编码和AI代理方面性能显著提升,在FrontierCode和DeepSWE基准上分别达到43.6%和65.3%,超越Claude Sonnet 5和GPT-5.6 Terra。API定价为每百万输入token 0.75美元,输出token 3.75美元,比前代便宜50%。

以上摘要由 AI 生成,可能存在误差。事实请以原文为准。

正文节选

Gemini 3.7 Flash lands with coding gains and undercuts its three-week-old predecessor's price by 50% Google has yet another Flash model. Just three weeks after Gemini 3.6 Flash, Google is already shipping Gemini 3.7 Flash. The company calls it its most capable workhorse model yet for coding and AI agents. Google credits "awesome algorithmic improvements" for the rapid jump in "intelligence". Code quality sees the biggest gains over its predecessor. On the FrontierCode benchmark, the model scores


发布时间:2026-08-14 02:41
抓取时间:2026-08-14 03:46
来源机构:THE DECODER
阅读原文the-decoder.com