Alibaba Group has achieved a major milestone in artificial intelligence, with its voice and coding models outperforming offerings from OpenAI and Google on key global benchmarks. The breakthrough marks a significant shift in the competitive AI landscape, as a Chinese company now leads in two critical technical domains previously dominated by U.S. firms.
Alibaba’s Fun-Realtime-TTS-Preview voice model secured fifth place on the Artificial Analysis Speech Arena leaderboard with a score of 1,190, making it the only Chinese-engineered voice system in the global top five. The model ranks ahead of voice AI from OpenAI and xAI. In the same organization’s word error rate index, Alibaba’s Fun-Realtime-ASR model took first place with an error rate of just 1.8 percent, meaning fewer than two words out of every 100 were transcribed incorrectly. The model also holds top domestic rankings across text-to-speech, automatic speech recognition, and end-to-end voice chat categories.
Tongyi Lab, Alibaba’s AI research division, has been developing speech capabilities for over a year. In 2025, the lab released its FunAudio-ASR model with advanced noise reduction that reduced hallucination rates from 78.5 percent to 10.7 percent.
In coding, Alibaba’s Qwen3.7-Max model scored 1,541 on the Code Arena leaderboard, ranking fourth globally as of May 26. The model surpassed GPT-5.5 and Gemini-3.5-Flash, with only different versions of Anthropic’s Claude models ranking higher. Alibaba is the only non-U.S. firm in the top five.
Qwen3.7-Max is designed for autonomous agent-based coding tasks and can operate without human intervention for up to 35 hours. It handles complex workflows including building front-end prototypes and managing large multi-file software projects, distinguishing it from conventional chatbots.
These dual achievements across voice and coding illustrate how Chinese AI labs are rapidly closing the gap with American counterparts on standardized benchmarks. Alibaba has invested heavily in AI infrastructure and committed roughly $53 billion to AI initiatives, positioning Qwen as a multi-modal platform competing across text, code, speech, and vision. The results signal growing global competition in AI development beyond the traditional U.S. dominance.
Read Article: NPCI Mandates Bank-Verified Recipient Names on UPI Apps to Curb Fraud
