[OpenAI Releases GPT-5.4 and GPT-5.4 Pro Models, Supporting Up to 1 Million Tokens in Context Window] Reported by Jinse Finance, OpenAI has officially launched GPT-5.4, featuring a 'Thinking' mode, available on ChatGPT, API, and Codex platforms. Additionally, GPT-5.4 Pro, optimized for complex tasks, has been released. OpenAI stated that in ChatGPT, GPT-5.4 demonstrates improved deep web research capabilities and better context retention during prolonged thinking processes—users can now interrupt the model during its response, add instructions, or adjust its direction. The guidance feature has been rolled out on Android and web platforms this week, with the iOS version coming soon. OpenAI developers claim that GPT-5.4 possesses native computer usage capabilities. Codex and API support up to 1 million context tokens, enabling optimal proxy coding for complex tasks, providing scalable tool searches within larger ecosystems, and delivering more efficient reasoning for long-term, tool-intensive workflows.