Google DeepMind 于 11 月 18 日正式发布 Gemini 4 Neon,定位介于 Carbon 与 Hydrogen 之间的中杯型号,主打更强的可控安全、更低价格与更快速度。这是谷歌在短短七周内推出的第四款模型。Google DeepMind released Gemini 4 Neon on Nov 18, a mid-tier model between Carbon and Hydrogen focused on safety, lower cost and higher speed — the company's fourth model in seven weeks.
Neon 支持文本、图像与视频多模态输入,上下文窗口达 100 万 token;在 Artificial Analysis 智能指数上得 61 分,位列全球第三,仅次于 GPT-6.2 Astra 与 Claude Fable 5.2。定价为每百万输入 token 3 美元、输出 15 美元,比 Carbon 标价低约 25%,推理速度约 110 tokens/s,接近 Carbon 的两倍。网络安全专用版 Neon for Cyber 将向受信任的防御人员开放。Neon handles text, images and video with a 1M-token context and scores 61 on the Artificial Analysis Intelligence Index (#3 worldwide, behind GPT-6.2 Astra and Claude Fable 5.2). Priced at $3/$15 per million input/output tokens, it is ~25% cheaper than Carbon and runs at ~110 tokens/s, nearly double Carbon. A Neon for Cyber variant will open to trusted defenders.