Unverified50% confidenceFactExact time
OpenAI o1 is a reasoning model released in September 2024, trained through reinforcement learning to develop a "slow thinking" capability.
1
Sources
50%
Confidence
Long-term
Relevance
9/1/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAndrew Ng and Colin Jarvis, OpenAI's Head of AI Solutions, jointly launched a short course called 'Reasoning with O1'.66% similarUnverifiedColin Jarvis emphasized in the course that O1's major long-term breakthrough lies in test-time / inference-time scaling.62% similarUnverifiedKimi-2.5是月之暗面（Moonshot AI）推出的强化学习推理模型，是国内首批明确以强化学习推理为核心训练范式的模型之一61% similarUnverified2020年OpenAI的Kaplan等人论文首次系统证明神经网络性能与模型参数量、训练数据量、计算量之间存在幂律关系61% similarUnverifiedTogetherAI 在 2024 年发表研究论文将集成学习思路引入大语言模型领域，实验表明将多个开源模型输出交由聚合模型综合处理后结果可超越任何单一模型61% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/836381API
curl https://kongchang.com/api/v1/knowledge/claims/836381MCP
get_claim(id=836381)