隨著中國公司急於從生成式人工智慧聊天機器人中賺錢，百度推出了 Ernie Bot 的付費版本

Baidu, a prominent internet search company in China, has initiated a paid version of its AI chatbot service, Ernie Bot. This move comes as Chinese tech firms are in a race to monetize their generative artificial intelligence technology. The professional plan provides users with access to the recently launched Ernie Bot 4. 0 at a monthly cost of 59. 9 yuan (US$8. 18), which is reduced to 49. 9 yuan for auto-renewal subscriptions. This information is available on the company's official website. The subscription permits users to make 100 inquiries every three hours via Ernie Bot 4. The more basic version, Ernie Bot 3. 5, remains free for public use.

Ernie Bot 4. 0 was unveiled by Baidu's co-founder and CEO, Robin Li Yanhong, last month. He promised improved capabilities in understanding complex questions, generating images, and handling basic arithmetic. 'Its generalized abilities are by no means inferior compared to GPT-4,' Li stated at its launch on October 17, referring to the latest ChatGPT model from Microsoft-backed OpenAI. The subscription fees for Ernie Bot are lower compared to OpenAI's ChatGPT Plus membership, which is priced at US$20 monthly and offers faster response times and priority access to new features.

Baidu's move is part of a larger trend among Chinese tech firms to monetize and commercialize their large language models (LLMs), which have surpassed 100 in number since OpenAI launched ChatGPT in November 2022. LLMs, the power behind these chatbots, are deep-learning AI algorithms that can recognize, summarize, translate, predict, and generate content using very large data sets. Some Chinese firms have already offered their LLMs to enterprise users for a fee calculated based on the number of texts processed and generated. Baidu's Qianfan, its model-as-a-service platform, offers various AI models and capabilities with prices starting from 0. 002 yuan per 1,000 tokens. Tokens refer to units of text that the model reads and generates, with their size varying among different models and languages. For instance, 1,000 tokens is about 750 words in English under OpenAI's scheme.