After December's preview, o3 brought private chain-of-thought reasoningStep-by-step logical thinking in AI models — chain-of-thought before answering hard problems. to production alongside the smaller o4-mini. Together they anchored OpenAIAn AI research company — created GPT, ChatGPT, DALL-E, and the o-series reasoning models.'s reasoningStep-by-step logical thinking in AI models — chain-of-thought before answering hard problems. tier until GPTGenerative Pre-trained Transformer — OpenAI's family of large language models from GPT-1 through GPT-5.-5 unified fast and thinking modes — and kept pressure on Google GeminiGoogle's family of multimodal AI models — successors to PaLM and competitors to GPT and Claude. 2.5 and AnthropicAn AI safety company — creators of the Claude family of language models. Claude 4.

o3 moved OpenAIAn AI research company — created GPT, ChatGPT, DALL-E, and the o-series reasoning models.'s second-generation reasoningStep-by-step logical thinking in AI models — chain-of-thought before answering hard problems. models from preview to production.

reasoningStep-by-step logical thinking in AI models — chain-of-thought before answering hard problems. plus tool use became the default stack for hard coding and research agents.