Gemini 4 Argon 开始付费 rollout
面向付费用户开放，强调推理与 Agent 编码能力及安全监控。
为什么重要
开发者可关注其付费可用性、安全机制和内部工程应用案例。
阅读原文arstechnica.com
AI 解读（根据原文生成）
Google 几周前披露 Gemini 4 Argon 后，现开始更广泛 rollout，但目前仅面向付费用户，低档计划也不能完全访问。文章称其能力没有变化，Google 仍引用上月发布的基准来证明高级推理和 agentic coding 能力，并提到模型经过训练会拒绝恶意指令，配有自动化 misalignment monitor。Google 还举例称 Argon 已在内部被数千名 Googlers 使用，用于 YouTube 评论系统、Project Wing 无人机物理模拟器和机器人项目。
要点（AI 据原文整理，数字均已与原文核对）
- Argon is beginning its wide rollout today, but it's only coming to paying users at this time.
- Nothing about Gemini 4 Argon's capabilities has changed.
- Google still points to the benchmarks it published late last month as evidence of the model's advanced reasoning and agentic coding capabilities.
- The model has allegedly been carefully trained to refuse malicious instructions.
- There is an automated misalignment monitor that can stop processing if it detects an unintended workflow.
- Argon identified redundant emoji rendering that, when eliminated, dropped server load by 45 percent.
