Ali Yun Tong Yiqian Ask 2.5 large model released, known as “a number of capabilities to catch up with GPT-4.

In the Ali Cloud AI Smart Leaders Summit – Beijing Station event, Tongyi thousand Question 2.5 large model was released, claiming a number of capabilities to catch up with GPT-4.

According to the official introduction of Ali Cloud, the Tongyi large model has exceeded 90,000 service enterprises through Ali Cloud, and the cumulative downloads of the Tongyi open source model have exceeded 7 million. The application process of Tongyi has accelerated, and it has now entered PC, mobile phone, automobile and other fields.

At the event site, Ali Cloud officially released Tongyi thousand Question 2.5, known as “ability upgrade, comprehensively catch up with GPT-4″, in the Chinese context, text understanding, text generation, knowledge & life advice, chat & dialogue and security risks and many other capabilities to catch up with GPT-4.

In addition, compared with Tongyi Qianwan 2.1, Tongyi Qianwan 2.5 has a number of ability improvements, including a 9% increase in understanding ability, a 16% increase in logical reasoning, a 19% increase in instruction compliance, and a 10% increase in code ability.

At the same time, Tongyi’s latest open source 110 billion parameter model has achieved the best results in multiple benchmark evaluations, surpassing Meta’s Llama-3-70B and declaring it to be “the most powerful model in the open source field.”