Alibaba's Qwen team has released Qwen3.8-Flash, a multimodal Mixture-of-Experts model that serves as an early preview of the upcoming Qwen4 architecture. This open-weight model is now accessible via the QwenCloud API, with pricing set at $0.16 per 1 million input tokens and $0.47 per 1 million output tokens for the production version. The model boasts 125 billion parameters and an additional 51 billion N-gram parameters. AI
IMPACT This release provides a new multimodal MoE model and an early look at the Qwen4 architecture, potentially influencing the development of similar models and offering competitive pricing for API access.
RANK_REASON Frontier-lab model release with system card. [lever_c_demoted from frontier_release: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →