Event ·Official first-hand

Terminal X 用 GPT-6.1 让评估 Agent 达到 88.6% 高级投资人一致率

Terminal X 推出 Eval Agent，用 54 项标准的评分规则来给模型输出打分。在 GPT-6.1 Sol 上运行后，它与资深投资人的判断一致率达到 88.6%。该系统还能多捕捉 25% 偏离基金投资指令的情况。

Current read

OpenAI 这篇案例讲 Terminal X 怎么用 GPT-6.1 做投资判断评估，54 项评分规则对齐资深投资人，做评估 Agent 的可以看看思路。

11 sources31° AI heatLast updated 2026/10/20 07:00:00

Evidence

11 sources

Primary sourceOpenAI Blog

2026/10/20 07:00:00View source

Related source 1Notion

2026/10/19 19:48:43View source

Related source 2@koltregaskes

2026/10/19 14:28:00View source

Related source 3lmarena.ai

2026/10/19 15:09:35View source

Related source 4@OpenAIDevs

2026/10/19 18:01:27View source

Related source 5IT之家

2026/10/20 06:56:15View source

Related source 6Rest of World

2026/10/20 14:43:02View source

Related source 7Decoder

2026/10/21 08:32:54View source

Related source 8theverge

2026/10/21 19:53:20View source

Related source 9rohanpaul_ai

2026/10/22 06:35:44View source

Related source 10Tibor Blaho

2026/10/18 10:57:33View 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.

Open the story