Gemini 4 Argon 瞄准更长、更复杂的 AI 任务

Google has announced Gemini 4 Argon, a new AI model designed for demanding professional work, including software engineering, research and cybersecurity. Introduced on September 30, the model is initially reaching selected security partners, with broader availability planned after additional testing. The emphasis is on maintaining useful performance through tasks involving many steps. A professional assignment might require examining documents, developing an approach, testing it and revising the result. Success depends on keeping those stages connected. For prospective customers, the announcement presents two separate questions: how well Argon performs and when they can use it. Early access does not mean the model is already available throughout the Gemini app or to every developer. …