ماذا تجيب هذه الصفحة
DeepSeek-V3.2 is a large language model designed to harmonize high computational efficiency with strong reasoning and agentic tool-use performance. It introduces DeepSeek Sparse Attention (DSA), a fine-grained sparse att
- DeepSeek · deepseek/deepseek-v3.2
- text->text · مسار نموذج الصين
- 163,840 context · 0.26 US$ input