यह पेज क्या बताता है
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 · China model route
- 1,63,840 context · $0.26 input