Alibaba

Alibaba Group says it was established in 1999 by 18 people led by Jack Ma, a former English teacher from Hangzhou, China, and now describes itself as "a global technology company focused on AI + Cloud and commerce."¹ Its AI services run on Qwen, its family of large language and multimodal models.¹ The Qwen team first published weights on August 3, 2023, when it released Qwen-7B and Qwen-7B-Chat on ModelScope and Hugging Face,² and the Qwen account on Hugging Face held 469 model repositories on October 13, 2026.³ Alibaba Cloud sells API access through Model Studio, which lists Qwen next to third-party models such as DeepSeek, GLM, Kimi and MiniMax, plus Alibaba's own image, video, speech and world models, including the Wan video line.⁴

Alibaba's models turn up across MetaPrompt's launch posts. In Cheapest LLM APIs per 1M Output Tokens, Alibaba Cloud's qwen3.8-flash lists the lowest output price of the 19 models ranked: $0.47 per 1M output tokens and $0.15 per 1M input tokens in the Singapore (International) deployment, against $6 output for the flagship qwen3.8-max.⁵ Hugging Face reports that Alibaba has more derivative models on the Hub than Google and Meta combined, with more than 113,000 built on the Qwen family, a spread covered in Chinese Open-Weight Models: Downloads, Derivatives and Token Share.⁶ For video, Model Studio lists wan3.0-video at up to 30 seconds per clip at up to 1080P,⁷ and the Wan technical report is the worked example in How AI Video Generation Works.⁸ In speech recognition, Qwen3-ASR-1.7B, released under Apache 2.0, averages a 5.40% word error rate on the Hugging Face Open ASR Leaderboard's October 9, 2026 version, the lowest average among the open-weight models in the top-20 table of What Word Error Rate Measures, and Why Speech-to-Text WERs Disagree.⁹¹⁰