Milan, 24 April (LaPresse) – DeepSeek has released its eagerly awaited next-generation artificial intelligence model, the open-source V4, which – according to the company – rivals the leading proprietary US models from firms such as OpenAI and Google DeepMind. This is reported by the *South China Morning Post*, which states that two versions have been launched. The V4-pro boasts 1.6 trillion parameters, making it – as the Chinese newspaper writes – the largest model ever created by the Hangzhou-based start-up in terms of the number of parameters, whilst the smaller model, the V4-flash, has 284 billion. Both models feature a context window of one million tokens, a key feature that determines the amount of information an AI system is capable of processing, and which DeepSeek claims to have achieved with ‘world-leading’ cost efficiency.