In an explanation of the company's Wechat account, the version of the “Qwen 2.5-Max” version, Microsoft, Openais “GPT-4O”, the roof company of Facebook Meta'nın Llama 3.1, and the V3 model from Depseek as well as better performance better performance better. Performance was transferred.
Deepseek, the Chinese company, led to a jerk in the sector that was dominated by US companies with the low -coast language model (LLM).
The new version of Alibaba's new version of the model on the first day of the spring festival in China, the success of Deepseek's success, not only American companies, but also under pressure to develop industry giants in China under the pressure of the development of artificial intelligence models.
When using the company together, QWEN 2.5-Max was more successful in information processing, programming, the comprehensive assessment of integrated capacities and compliance with human preferences than the open source counterparts and showed a close performance for the Claude supported by Amazon 3.5-lonnet.
The Chinese company Deepseek's “Deepseek V3”, which was developed by a low budget, fell better than the colleagues developed in the USA with larger investments and led to the depreciation of the stocks of American technology companies.
Bir yanıt yazın