News

Alibaba Launches Low-Cost AI Translation Model Supporting 92 Languages

• By John K. Waters
• 07/25/2025

Chinese multinational technology company Alibaba announced the release of Qwen-MT, a machine translation model that supports 92 languages and offers significantly reduced costs compared to existing solutions, as the Chinese tech giant seeks to compete in the artificial intelligence translation market.

The new model, built on Alibaba's Qwen3 foundation and available through the company's API service, uses a lightweight Mixture of Experts architecture to deliver translations at $0.5 per million output tokens - substantially lower than current market rates for comparable AI translation services.

Qwen-MT covers languages spoken by over 95% of the global population, including major languages like English, Chinese, Spanish, and Arabic, as well as less common dialects such as Pangasinan and Maithili. The model incorporates reinforcement learning techniques trained on what Alibaba describes as "trillions of multilingual tokens."

In benchmark testing, Alibaba said Qwen-MT outperformed similarly-sized models, including OpenAI's GPT-4.1-mini and Google's Gemini-2.5-Flash on Chinese-English and English-German translation tasks, while maintaining competitive performance against larger models like GPT-4.1 and Gemini-2.5-Pro.

The company conducted human evaluations across 10 major languages using three independent professional translators for each test sample. According to Alibaba, the results showed Qwen-MT achieved higher acceptance and excellence rates compared to competing models, though the company did not provide specific comparative data or identify which competitors were included in the evaluation.

The model offers enterprise-focused features including terminology control, domain-specific prompting, and translation memory integration. These capabilities allow businesses to maintain consistent translation of technical terms and adapt output style for different contexts, from legal documents to social media content.

"The model aims to provide global users with a smart, flexible, and efficient translation experience through APIs," Alibaba stated in its announcement, positioning the service for high-concurrency environments and latency-sensitive applications.

The launch comes as major technology companies intensify competition in AI-powered language services. Google Translate supports over 100 languages, while Microsoft's Azure Translator covers approximately 90 languages. Amazon Web Services offers translation services for 75 languages.

Qwen-MT's pricing structure could appeal to startups and applications requiring heavy localization, particularly in e-commerce, content management, and customer service sectors, where translation costs can significantly impact operational expenses.

The model is immediately available through Alibaba's Qwen API platform, supporting both real-time translation requests and batch processing for high-volume applications.

Alibaba has positioned artificial intelligence as a key growth area as it faces increasing competition in its core e-commerce business and seeks to expand its cloud computing services globally.