OpenAI 发布 GPT-6.1 Terra，在 Artificial Analysis 评测中提升了成本效率

发生了什么

OpenAI 已发布 GPT-6.1 Terra。Artificial Analysis 报告称，在最高推理强度下，该模型每项任务的成本不到 GPT-5.6 Terra 的三分之一，同时具备更高的智能水平（0.40 美元，对比 1.40 美元）。 输入/输出定价为每百万 token 1 美元/5 美元，之前为 2 美元/12 美元。Artificial Analysis 报告称，在最高推理强度下，Intelligence Index 提升了 6 分；在 AA-Omniscience 上，幻觉率从 88% 降至 38%，准确率则提高了 2 个百分点。低和中推理强度设置进入了 Intelligence Index 的成本效率前沿。

接下来读什么

• 今日还有哪些重要变化Nvidia 开源 Flora：机器人视频标注成本预计最高比 GPT-6 Astra 低 125 倍

这件事到目前为止

OpenAI has released GPT-6.1 Terra, which it says can interpret screens and take action across apps, with access now available through the API and to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, according to The Decoder. API pricing is $1 per million input tokens and $5 per million output tokens, down from GPT-5.6 Terra’s $2 and $12, according to Artificial Analysis; The Decoder reports cached input pricing of $0.05 per million tokens. In Artificial Analysis evaluations at maximum reasoning effort, Terra gained six points on the Intelligence Index over its predecessor while cost per task fell from $1.40 to $0.40. On AA-Omniscience, its hallucination rate fell from 88% to 38%, alongside a two-point accuracy increase. Terra scored 48 on the Coding Agent Index v2.0, below GPT-6.1 Sol’s 54 and above GPT-6 Luna’s 31. OpenAI says Terra completes tasks more than three times faster on average than Astra and Sol, while continuing to recommend those models for deep research and complex coding, according to Baoyu. Greg Brockman said Terra’s computer use feels faster than his own unaided computer use.