Unverified50% confidenceFactExact time
OpenAI's o1 series and Google's Gemini 2.5 series internalize chain-of-thought as the model's default behavior, with models trained to automatically perform internal reasoning.
1
Sources
50%
Confidence
Medium-term (~90 days)
Relevance
8/29/2026
First Seen
Valid until: 11/27/2026
Sources
Related Entities
Related Claims
UnverifiedOpenAI's o1 model was released in September 2024; through reinforcement learning it was trained to perform internal reasoning before outputting a final answer, with intermediate reasoning steps existing as special tokens that are invisible to users.71% similarUnverifiedOpenAI 推出了 o1、o3 系列推理模型，Anthropic 在 Claude 中引入了 extended thinking 功能，Google Gemini 也提供了类似能力70% similarUnverifiedThe training frameworks for models such as OpenAI's GPT series, Meta's LLaMA, and Google's Gemini are all based on PyTorch or JAX (Python interface).69% similarUnverifiedOpenAI的o1模型仅展示经过处理的'思考摘要'而非原始推理token，而谷歌AI Studio则提供相对更完整的内部推理可见性69% similarUnverifiedOpenAI models use the Codex environment in Frontier Code testing, Gemini models use Gemini CLI, and open-source models use Mini Agent.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/819694API
curl https://kongchang.com/api/v1/knowledge/claims/819694MCP
get_claim(id=819694)