Event ·Multi-source

Agent Arena 数据:Gemini 4 Carbon 成本扩展性优于竞品

Agent Arena 发布的模型成本对比显示,Google 的 Gemini 4 Carbon 在任务复杂度上升时的相对成本优势扩大。对于典型任务,OpenAI 的 GPT-6 Astra 成本是 Carbon 的 3.4 倍,Anthropic 的 Claude Opus 5.5 是 3.9 倍,GPT-6.1 Sol 则约为 Carbon 的 0.54 倍。在最贵的 5% 会话中,Astra 相对成本约翻倍,Opus 升至 4.2 倍,Sol 的折扣收窄。在高成本会话区,Gemini 4 Argon 和 Xiaomi 的 MiMo-V2.6 Flash、Pro 也在排行榜上抢占名额。过去两周有超过 1,000 个会话成本超过 100 美元,模型选择的差异对最终账单影响可观。

Current read

Arena 拿真实智能体会话算了一笔账:任务越难,Gemini 4 Carbon 越省钱,Astra 和 Opus 5.5 贵出三四倍,选错模型账单差很多。

11 sources49° AI heatLast updated 2026/11/4 17:46:04

Evidence

11 sources
Related source 1rohanpaul_ai
View source
Related source 2Artificial Analysis
View source
Related source 4TestingCatalog
View source
Related source 5shao__meng
View source
Related source 6The Rundown AI
View source
Related source 8Nathan Lambert
View source
Related source 9Simon Willison’s Weblog
View source

Conflict check

Dedup data only shows that these sources discuss the same event, not that they agree. There is no structured supporting or refuting evidence yet, so no conflict is inferred automatically.