Event ·Official first-hand

GPT-6.1 Sol 减少事实错误响应

OpenAI 发布 GPT-6.1 Sol，在故意设计的事实性提示测试中，相比 GPT-6 Sol，在低推理努力下减少了约 32% 的错误响应。在自动化安全审查评估中，未观察到尝试绕过安全审查的行为。该模型在事实准确性方面有明显改进。

Current read

OpenAI 新版 GPT-6.1 Sol 事实错误减少 32%，安全审查未被绕过，比前代更可靠。

11 sources83° AI heatLast updated 2026/9/29 17:53:58

Evidence

11 sources

Primary source@OpenAIDevs

2026/9/29 17:53:58View source

Related source 1OpenAI

2026/9/29 17:26:17View source

Related source 2rohanpaul_ai

2026/9/29 17:33:42View source

Related source 3Techsauce

2026/9/30 06:57:31View source

Related source 4Blognone

2026/9/30 11:11:14View source

Related source 5ARC Prize

2026/9/30 20:18:27View source

Related source 6IT之家

2026/9/28 01:10:40View source

Related source 7lmarena.ai

2026/9/28 15:57:26View source

Related source 8Decoder

2026/9/28 18:02:10View source

Related source 9techcrunch

2026/9/29 17:15:00View source

Related source 10PolymarketMoney

2026/9/29 17:18:05View 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