Qwen3-VL-8B-Thinking
Qwen3-VL-8B-Thinking is an AI model published by 阿里巴巴, released on 2025-10-15, for 推理大模型, with 87.7B parameters, and 256K tokens context length, under the Apache 2.0 license.
Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology
Qwen3-VL-8B-Thinking 为 Qwen3-VL 系列的中型推理增强多模态模型,于 2025-10-15 与 4B/8B Instruct/Thinking 同步发布。
支持图像/视频输入与文本输出,覆盖 OCR(32 语种)、版面解析、复杂空间推理与长视频时序定位,以及“视觉代理”场景(GUI 元素识别、功能理解与步骤执行)。
欢迎关注 DataLearner 官方微信,获得最新 AI 技术推送
