Alibaba releases Qwen3.8-Flash-Next, targeting "ultimate cost efficiency"

Alibaba's Qwen team has unveiled the Qwen3.8-Flash-Next model, a mixture-of-experts architecture that activates only 6 out of 125 billion parameters per token. This model boasts a training cost that is one-ninth that of its larger competitors, such as DeepSeek-V4-Flash and Claude Opus 4.6, adding pricing pressure on major players like OpenAI and Anthropic. This development marks a strategic move for Alibaba in the AI market, as the new model aims for ultimate cost efficiency, potentially influencing purchasing decisions among large enterprises in the sector.