Media: "quietly shifting", Indian enterprises favor Chinese AI large models

Recently, an increasing number of Indian technology companies turning to Chinese artificial intelligence large models has attracted the attention of the Indian media. "Quiet shift: Indian developers favor Chinese AI." The Indian newspaper "Telugu Times" recently reported with this title, stating that a more pragmatic story is unfolding within Indian technology companies: facing the high cost of American AI services, Indian developers are quietly turning to models developed in China, which are more cost-effective and openly available. The "Nikkei Asian Review" quoted Future Assets Venture Capital India CEO Puneet Kumar in its report on the 13th, stating that since the mid-2025s, many Indian technology startups he has been in touch with have been using Chinese open-source large language models like DeepSeek and Qianwen, which have helped them reduce costs by "an order of magnitude." The report cited a Chinese AI large model as an example, with an input price ranging from $0.19 to $1.74 per million tokens and an output price ranging from $0.51 to $5.40 per million tokens. In comparison, the input price for OpenAI GPT-5.5 is between $5 and $12 per million tokens, with an output price ranging from $30 to $54 per million tokens.