Alibaba’s AI coding model Qwen 3 Coder soars in popularity, challenging Claude Sonnet 4
The model’s usage share on AI marketplace OpenRouter hit 20 per cent as of mid-August, behind only Anthropic’s coding model
Alibaba Group Holding’s artificial intelligence coding model Qwen 3 Coder has been gaining popularity globally amid intensified competition in the AI-assisted coding sector, according to data from a third-party agency.
The model’s usage share on AI marketplace OpenRouter soared to more than 20 per cent as of mid-August, trailing only Amazon.com-backed Anthropic’s Claude Sonnet 4 with 31 per cent of usage.
Alibaba’s Qwen team released the Qwen 3 Coder on July 23, claiming it delivers top-tier AI coding performance on par with Anthropic’s Claude Sonnet 4, citing various third-party benchmark tests. Claude Sonnet 4 is widely regarded as the industry’s leading coding model. Alibaba owns the Post.
Qwen 3 Coder’s performance boost was due to expanded data sets that include a high portion of coding-related data for training, improved overall data quality and the use of large-scale reinforcement learning, according to the Qwen team.
Alibaba’s coding tool launched at a time when the AI-assisted coding sector was becoming increasingly crowded with more players entering the space.
A slew of Chinese tech majors – from ByteDance to Tencent Holdings and Baidu – have all released AI coding tools to ride on the so-called vibe coding wave, which is the trend of using AI to help generate, complete and debug code.