qwen3-4b · Qwen
Achieves effective integration of thinking and non-thinking modes, allowing mode switching during conversations. Its reasoning ability reaches state-of-the-art (SOTA) levels among models of the same scale, with significantly enhanced human preference alignment. There are notable improvements in creative writing, role-playing, multi-turn dialogue, and instruction following, resulting in a noticeably better user experience.
Achieves effective integration of thinking and non-thinking modes, allowing mode switching during conversations. Its reasoning ability reaches state-of-the-art (SOTA) levels among models of the same scale, with significantly enhanced human preference alignment. There are notable improvements in creative writing, role-playing, multi-turn dialogue, and instruction following, resulting in a noticeably better user experience.
On AIHubMix, Qwen3 4B costs $0.046 per million input tokens and $0.46 per million output tokens.
Qwen3 4B is available through the AIHubMix unified API. The API is OpenAI-compatible: point your OpenAI SDK at https://aihubmix.com/v1, use your AIHubMix API key, and set the model name to qwen3-4b — no other code changes needed.
Qwen3 4B is developed by Qwen. AIHubMix aggregates it alongside models from other providers behind one API and one bill.
Qwen3.8 Omni Flash is Alibaba Cloud Qwen's next-generation native multimodal model…
Alibaba Cloud has launched the decision model decision-model-preview. This structured…
Qwen3.8-27b is an Alibaba-released dense vision-language model with open-source weights…
Qwen3.8-Max-0902 (also known as qwen3.8-max-2026-09-02) is a snapshot version of Alibaba…
Qwen3.8 Flash is Alibaba Cloud Qwen’s flagship native vision-language model for coding…
Wan 3.0 (Tongyi Wanxiang 3.0) is an integrated video generation and editing model…
Use Qwen3 4B via the AIHubMix unified API — one interface for every major LLM.