Alibaba’s artificial intelligence unit Qwen launched the Qwen3.8-Flash model on Wednesday, positioning it as a cost-efficient upgrade with a 90% reduction in training costs versus its predecessor Qwen3.7-Plus.
The new model maintains competitive performance in programming and office automation tasks while introducing a standard context window of 262,144 tokens, expandable to 1 million tokens. This capacity supports handling large documents, extended conversations, and complex research inputs without degradation in response quality.
Alibaba priced API access at 1 yuan (US$0.1488) per million input tokens and 3 yuan per million output tokens. The company also released open-source weights for Qwen3.8-Flash-Next, a prototype architecture intended as a foundation for the upcoming Qwen4 family of models.
Qwen’s models remain among the most widely adopted in China, where competition in the AI sector continues to intensify. The launch follows Alibaba’s strategy to balance performance gains with operational efficiency amid rising computational demands across enterprise and developer use cases.
