Tongyi Qianwen’s Flagship Model Qwen-Long Announces A 97% Price Reduction

May 21, 2024, 14:37pm2024/05/21 14:38:03 Pandaily

ByteDance officially released the Doubao large model on May 15th, announcing that the price of the large model has entered the “cent era,” claiming to be “99.3% cheaper than industry standards.

Today, Alibaba Cloud followed suit and dropped a bombshell: the flagship model Qwen-Long of the GPT-4 level announced a price reduction. The API input price decreased from 0.02 yuan/thousand tokens to 0.0005 yuan/thousand tokens, a decrease of 97%.

Simply put, 1 yuan can buy 2 million tokens, equivalent to the amount of text in 5 “Xinhua Dictionary”. This model supports a maximum of 10 million tokens for long-text input, and after the price reduction it is about 1/400 of the price of GPT-4.

Alibaba<img decoding="async" class="trade-card__loading" layout="fixed" height="50" width="50" src="https://assets.pandaily.com/uploads/2021/12/loading.png" placeholder /> Cloud officially launched the new Qwen 2.5 large model on May 9 this year, claiming to surpass GPT-4 in multiple capabilities.

According to Alibaba<img decoding="async" class="trade-card__loading" layout="fixed" height="50" width="50" src="https://assets.pandaily.com/uploads/2021/12/loading.png" placeholder /> Cloud’s official introduction, the Qwen large model has served over 90,000 enterprises through Alibaba<img decoding="async" class="trade-card__loading" layout="fixed" height="50" width="50" src="https://assets.pandaily.com/uploads/2021/12/loading.png" placeholder /> Cloud services, and the cumulative download volume of the open-source Qwen model has exceeded 7 million. The application process of Qwen is accelerating and has now entered fields such as PC, mobile phones, and automobiles.

At the event, Alibaba<img decoding="async" class="trade-card__loading" layout="fixed" height="50" width="50" src="https://assets.pandaily.com/uploads/2021/12/loading.png" placeholder /> Cloud officially released Qwen 2.5, claiming to ‘upgrade capabilities and surpass GPT-4 comprehensively.’ In the Chinese context, it surpasses GPT-4 in multiple abilities such as text understanding, text generation, knowledge question answering & life advice, chat & dialogue, and security risks.

In addition, compared to Qwen 2.1, Qwen 2.5 has seen improvements in multiple abilities: a 9% increase in comprehension ability, a 16% increase in logical reasoning, a 19% increase in instruction compliance, and a 10% improvement in coding ability.

At the same time, the latest open-source model with 110 billion parameters from Qwen achieved top scores in various benchmark evaluations and surpassed Meta’s Llama-3-70B model to become ‘the most powerful model in the open-source field.’

Subscribe now to get unlimited access.